In This paper we introduce the integration of Renormalization Group (RG) methods with Deep Q-Networks (DQNs) to improve reinforcement learning in high-dimensional state spaces. RG methods provide multi-scale analysis, enhancing state representation, learning stability, and exploration. The proposed RG-DQN algorithm uses hierarchical Q-value estimation and multi-scale representations, demonstrating superior performance on synthetic genomic data compared to traditional DQNs.}

Multimodal emotion classification (MEC) involves analyzing and identifying human emotions by integrating data from multiple sources, such as audio, video, and text. This approach leverages the complementary strengths of each modality to enhance the accuracy and robustness of emotion recognition systems. However, one significant challenge is effectively integrating these diverse data sources, each with unique characteristics and levels of noise. Additionally, the scarcity of large, annotated multimodal datasets in Bangla limits the training and evaluation of models. In this work, we unveiled a pioneering multimodal Bangla dataset, MAViT-Bangla (Multimodal Audio Video Text Bangla dataset). This dataset, comprising 1002 samples across audio, video, and text modalities, is a unique resource for emotion recognition studies in the Bangla language. It features emotional categories such as anger, fear, joy, and sadness, providing a comprehensive platform for research. Additionally, we developed a framework for emotion recognition that employs a cross-modal attention mechanism among unimodal features. This mechanism fosters the interaction and fusion of features from different modalities, enhancing the model’s ability to capture nuanced emotional cues. The effectiveness of this approach was demonstrated by achieving an F1-score of 0.64, a significant improvement over unimodal methods.

Physics-informed neural networks can be trained to serve as surrogate models by learning how spatial physical fields develop with change of parameters. This is an advantage over traditional numerical methods that can only be used to calculate a single solution field for a single discrete setting of a parameter. While many studies report accurate results of physics-informed deep learning of laminar flows, evaluations of surrogate modeling of high Reynolds number flows remain sparse. To contribute to this question, we here explore the capabilities and limits of physics-informed deep learning to solve the Reynolds-averaged Navier-Stokes equations for the flow around two airfoils under variable angles of attack. No labeled training data were provided and a single network was trained to simultaneously learn several solution fields in a single training run. The network captured essential flow features such as boundary layers and high and low pressure regions. Moreover, the qualitative correlations of lift and drag with the angle of attack and the higher drag of the thicker airfoil were captured. However, comparisons with reference simulations revealed an underestimation of the shear layer gradients as well as deviations of the fields' extrema. Thus, we also debate potential future improvements to increase the accuracy of the predicted flow. This work offers a prospect on the capabilities and current limitations of physics-informed neural networks for surrogate modeling of high Reynolds number flows. We argue that the method is promising and this work is aimed to encourage further research and development of the method when applied to Reynolds-averaged flows at high Reynolds numbers.

This study investigates the impact of AI regulations and adoption on labor markets and employment in the USA. In order to achieve the study's objective, a robust econometric approach was employed using Principal Components Analysis (PCA) to address multicollinearity among the variables. The regression analysis incorporated principal components representing overall AI adoption and education (PC1), AI innovation and academic output (PC2), economic growth (PC3), and unemployment and AI strategies (PC4). The analysis utilized heteroskedasticity-robust standard errors to ensure reliable coefficient estimates and tested for stationarity to confirm the stability of the time series data. Data were collected from 2010 to 2022, encompassing key AI-related and economic indicators. The results reveal that higher AI adoption and education levels initially lead to job displacement, negatively impacting labor market metrics. Similarly, AI innovation and economic growth driven by AI do not immediately translate into job creation, reflecting transitional challenges. However, the strategic implementation of AI significantly mitigates these adverse effects and enhances employment conditions. The study finds that comprehensive AI strategies, robust regulatory frameworks, and effective workforce retraining and upskilling programs are essential for supporting labor market stability and promoting employment growth. Based on the findings, the study recommended that there is the need to develop and implement comprehensive AI strategies that include robust regulatory frameworks to support workforce transitions. Policies should focus on retraining and upskilling programs to help displaced workers adapt to new AI-driven roles.

This paper explores the application of machine learning in financial time series analysis, focusing on predicting trends in financial enterprise stocks and economic data. It begins by distinguishing stocks from stocks and elucidates risk management strategies in the stock market. Traditional statistical methods such as ARIMA and exponential smoothing are discussed in terms of their advantages and limitations in economic forecasting. Subsequently, the effectiveness of machine learning techniques, particularly LSTM and CNN-BiLSTM hybrid models, in financial market prediction is detailed, highlighting their capability to capture nonlinear patterns in dynamic markets. The study demonstrates the advancements in predictive accuracy and robustness achieved by deep learning methods through empirical analysis and model validation. The findings contribute significantly to academic discourse and offer practical insights for investors, financial analysts, and policymakers navigating market volatility and optimizing investment strategies. Finally, the paper outlines prospects for machine learning in financial forecasting, laying a theoretical foundation and methodological framework for achieving more precise and reliable economic predictions.

This survey provides a comprehensive overview of in-context learning (ICL) in large language models (LLMs), a phenomenon where models can adapt to new tasks without parameter updates by leveraging task-relevant information within the input context. We explore the definition and mechanisms of ICL, investigate the factors contributing to its emergence, and discuss strategies for optimizing and effectively utilizing ICL in various applications. Through a systematic review of recent literature, we first clarify what ICL is, distinguishing it from traditional fine-tuning approaches and highlighting its unique characteristics. We then delve into the underlying causes of ICL, examining theories ranging from implicit meta-learning during pre-training to the emergence of task vectors in LLMs. The survey also covers various approaches to enhance ICL performance, including prompt engineering techniques, demonstration selection strategies, and methods for improving generalization across diverse tasks. Additionally, we discuss the limitations and challenges of ICL, such as its sensitivity to demonstration ordering and potential biases. By synthesizing findings from numerous studies, we aim to provide researchers and practitioners with a clear understanding of the current state of ICL research, its practical implications, and promising directions for future investigation. This survey serves as a valuable resource for those seeking to leverage ICL capabilities in LLMs and contributes to the ongoing discourse on the remarkable adaptability of these models.

This research addresses the challenge of integrating fairness considerations into optimization problems, focusing on the development of multi-objective optimization techniques that balance traditional efficiency metrics with fairness constraints. In many algorithmic decision-making processes, optimizing for efficiency alone can inadvertently perpetuate or exacerbate biases, leading to inequitable outcomes. This study explores how data analysis and machine learning can be leveraged to identify and mitigate these biases, ensuring that optimization models yield both efficient and fair solutions. The research presents a framework for incorporating fairness constraints into multi-objective optimization, utilizing various fairness metrics such as demographic parity, equal opportunity, and disparate impact. It investigates the application of this framework across different domains, including resource allocation in healthcare systems, loan approvals in finance, and personalized learning platforms in education. By analyzing real-world data sets, the study demonstrates how fairness-aware optimization can lead to more equitable outcomes without significantly compromising efficiency. Empirical results from case studies show that multi-objective optimization with fairness constraints can effectively balance the trade-off between efficiency and equity. The research also discusses the computational challenges and ethical considerations associated with implementing fairness constraints in optimization models. Strategies for addressing these challenges, such as regularization techniques and fairness-aware machine learning algorithms, are evaluated and presented. Findings highlight the importance of incorporating fairness into optimization processes to prevent algorithmic biases and promote social equity. The study concludes with practical recommendations for policymakers, practitioners, and researchers on adopting fairness-aware optimization techniques to ensure that algorithmic decision-making processes are both efficient and just.

Large Language Models (LLMs) have demonstrated promising capabilities to solve complex tasks in critical sectors such as healthcare. However, LLMs are limited by their training data which is often outdated, the tendency to generate inaccurate ("hallucinated") content and a lack of transparency in the content they generate. To address these limitations, retrieval augmented generation (RAG) grounds the responses of LLMs by exposing them to external knowledge sources. However, in the healthcare domain there is currently a lack of systematic understanding of which datasets, RAG methodologies and evaluation frameworks are available. This review aims to bridge this gap by assessing RAG-based approaches employed by LLMs in healthcare, focusing on the different steps of retrieval, augmentation and generation. Additionally, we identify the limitations, strengths and gaps in the existing literature. Our synthesis shows that proprietary models such as GPT-3.5/4 are the most used for RAG applications in healthcare. Also, there is a lack of standardized evaluation frameworks for RAG-based applications. In addition, the majority of the studies do not assess or address ethical considerations related to RAG in healthcare. It is important to account for ethical challenges that are inherent when AI systems are implemented in the clinical setting. Lastly, we highlight the need for further research and development to ensure responsible and effective adoption of RAG in the medical domain.

This paper presents a repository of image databases where the image features are augmented with embedded vector field (VF) features to enhance the machine learning (ML) and increase classification statistics. Six VFs are formulated and used for the purpose of embedding. Three of them generate real shaped singular points (SPs): springing, sinking and saddle. The other three VFs generate seven kinds of SPs which include the real shaped SPs and four complex shaped SPs: repealing and attracting (out and in) spirals; clockwise and counterclockwise orbits (center). Also, the paper describes the VF singularities, their shapes, and locations according to the image objects. Then, the present work defines the mappings between the SPs’ shapes if the VFs are separately embedded into the same image. Thus, the paper infers which VFs are most appropriate for embedding into a particular image database such that the augmented SPs shapes enhance the ML classification. Examples of images with different embedded VFs are shown, in the text, to support and validate the theoretical conclusions. The contributions of the paper are: a description of the SPs location in an image; the mappings between the SPs of the different VFs; the definition of an imprint of an image in a VF; the generation of databases’ imprints. The advantage of classifying an image database with embedded VFs is that the new database enhances ML and improves the classification statistics if compared to the classification of the original image database with the same ML classifier. Furthermore, images with embedded SPs create a useful augmentation to the original image databases.

This paper investigates the use of Physics-Informed Neural Networks (PINNs) in active learning cycles. We defined two scenarios: one semi-supervised and the other fully supervised. PINNs emphasize the integration of physical laws into neural networks to improve the predictive performance of vanilla neural networks and to enhance the efficiency of traditional methods for solving partial differential equations (PDEs). Key contributions include adapting existing computational frameworks to enable the use of Graph Neural Networks for solving problems that require the calculation of gradients on unstructured triangle meshes, a query strategy focusing on the physical loss, and a comparative analysis of this strategy against random sampling across both defined scenarios. This work establishes a foundation for future research aimed at expanding the application of Physics-Informed Graph Neural Networks (PIGNN) using active learning and addressing real-world problems in fluid dynamics and electrodynamics.

This study explores the limitations of current AI systems, which predominantly function in the digital realm, and investigates the transformative potential of embodied AI. Embodied AI involves AI agents equipped with physical bodies, enabling them to interact directly with the physical world. This research focuses on how embodied AI can bridge the "embodiment gap" and enhance human-machine collaboration and learning in dynamic environments. The research examines the fundamental differences between traditional AI systems and embodied AI, emphasizing the importance of physical interaction for contextual understanding, adaptive learning, and intuitive human-machine collaboration. It explores various applications of embodied AI, including robotics, autonomous vehicles, and assistive technologies, demonstrating how physical embodiment can improve performance, safety, and user experience. Through experimental studies and real-world case analyses, the study highlights the advantages of embodied AI in tasks requiring situational awareness, dexterity, and real-time decision-making. It also addresses the challenges associated with developing and deploying embodied AI systems, such as sensor integration, real-time processing, and human-machine interface design. Findings indicate that embodied AI can significantly improve the efficacy of AI systems in dynamic and unpredictable environments. By leveraging physical embodiment, AI agents can better understand and respond to their surroundings, facilitating more natural and effective interactions with humans. The research concludes with recommendations for advancing embodied AI, including interdisciplinary collaboration, investment in sensor and actuator technologies, and the development of standardized frameworks for embodied intelligence.

Accurate segmentation of skin lesions from dermoscopic images is crucial for early skin cancer detection, yet variations in lesion appearance and image artifacts present challenges. This study proposes two deep learning models, UCM-NetV2 and BNN-UCM-NetV2, to improve accuracy and computational efficiency. UCM-NetV2 enhances the UCM-Net architecture with a novel "cyber-structure" combining Multilayer Perceptron and CNN layers, improving prediction accu-racy without increasing the model's parameter count. BNN-UCM-NetV2 further optimizes for mobile deployment by binarizing activations and weights into 1-bit values, reducing computa-tional complexity and model size. Evaluations on the ISIC2017 and ISIC2018 datasets demonstrate that UCM-NetV2 outperforms existing methods in both accuracy and computational efficiency, achieving this with less than 0.04 GFLOPs. These models have the potential to make skin lesion analysis accessible to a broader population, particularly in resource-limited settings, allowing for proactive skin health monitoring and facilitating teledermatology. To foster further innovation in mobile health diagnostics, the source code for UCM-NetV2 will be available on GitHub: https://github.com/chunyuyuan/UCMV2-Net.

The integration of Artificial Intelligence (AI) into education has the potential to revolutionize traditional teaching and learning methods. AI can offer personalized learning experiences, streamline administrative tasks, enhance feedback mechanisms, and provide robust data analysis. Numerous studies have demonstrated the positive impact of AI on both student outcomes and teacher efficiency. However, caution must be exercised when implementing AI in education, considering potential risks and ethical dilemmas. It is essential to use AI as a tool to support human educators rather than replace them entirely. The adoption of AI in education holds the promise of creating more inclusive and effective learning environments, catering to students of diverse backgrounds and abilities. As AI technology continues to advance, the education sector can anticipate even more innovative applications, further shaping the future of learning. This abstract provides an overview of the multifaceted landscape of AI in education, highlighting its potential benefits, associated challenges, and the importance of responsible integration.

Incorporating knowledge graphs as auxiliary information to enhance recommendation systems can improve the representations learning of users and items. Recommendation methods based on knowledge graph can introduce user-item interaction learning into the item graph, focusing only on learning the node vector representations within a single graph; alternatively, they can treat user-item interactions and item graphs as two separate graphs and learn from each graph individually. Learning from two graphs has natural advantages in exploring original information and interaction information, but faces two main challenges: 1) in complex graph connection scenarios, how to adequately mine the self-information of each graph, and 2) how to merge interaction information from the two graphs while ensuring that user-item interaction information predominates. Existing methods do not thoroughly explore the simultaneous mining of self-information from both graphs and effective interaction information, leading to the loss of valuable insights. Considering the success of contrastive learning in mining self-information and auxiliary information, this paper proposes a dual-graph contrastive learning recommendation method based on knowledge graphs (KGDC) to explore a more accurate representations of users and items in recommendation systems based on external knowledge graphs. In the learning process within the self-graph, KGDC has strengthened and represented the information of different connecting edges in both graphs, and extracted the existing information more fully. In interactive information learning, KGDC reinforces the interaction relationship between users and items in the external knowledge graph, realizing the leading role of the main task. We have conducted a series of experiments on three standard datasets, and the results show that the proposed method can achieve better results.

This study involves the development of a hybrid forecasting framework that integrates two different models in a framework to improve prediction capability. Although the concept of hybridization is not a new issue in forecasting, our approach presents a new structure that combines two standard simple forecasting models uniquely for superior performance. Hybridization is significant for complex data sets with multiple patterns. Such data sets do not respond well to simple models, and hybrid models based on the integration of various forecasting tools often lead to better forecasting performance. The proposed architecture includes serially connected ARIMA and ANN models. The original data set is first processed by ARIMA. The error (i.e., residuals) of the ARIMA is sent to the ANN for secondary processing. Between these two models, there is a classification mechanism where the raw output of the ARIMA is categorized into three groups before they are sent to the secondary model. The algorithm is tested on well-known forecasting cases from the literature. The proposed model performs better than existing methods in most cases.

Digital recruitment systems have revolutionized the hiring paradigm, imparting exceptional efficiencies and extending the reach for both employers and job seekers. This investigation scrutinizes the efficacy of classical machine learning methodologies alongside advanced large language models (LLMs) in aligning resumes with job categories. Traditional matching techniques, such as Logistic Regression, Decision Trees, Naïve Bayes, and Support Vector Machines, are constrained by the necessity of manual feature extraction, limited feature representation, and performance degradation, particularly as dataset size escalates, rendering them less suitable for large-scale applications. Conversely, LLMs like GPT-4, GPT-3, and LLAMA, adeptly process unstructured textual content, capturing nuanced language and context with greater precision. We evaluate these methodologies utilizing two datasets comprising resumes and job descriptions to ascertain accuracy, efficiency, and scalability. Our results reveal that while conventional models excel with structured data, LLMs significantly enhance the interpretation and matching of intricate textual information. This study highlights the transformative potential of LLMs in recruitment, offering insights into their application and future research avenues.

Dams significantly impact the environment, industry, residential areas, and agriculture. Efficient dam management can mitigate negative impacts and enhance benefits such as flood and drought reduction, energy efficiency, water access, and improved irrigation. This study tackles the critical issue of predicting dam occupancy levels precisely within the framework of Integrated Water Resource Management (IWRM). Our research proposes that combining the physical model of evapotranspiration using the Penman–Monteith equation with data-driven models based on historical reservoir data, weather data, and consumption data is essential for accurately predicting occupancy levels. We implemented various prediction models, including Random Forest, Extra Trees, Long Short-Term Memory, Orthogonal Matching Pursuit CV, and Lasso Lars CV. To strengthen our proposed model with robust evidence, we conducted a statistical test on the MAPE values. We achieved a remarkable accuracy in predicting the occupancy level one month ahead, with an error margin of just 1% using Extra Trees. This study represents a pioneering effort in providing guidance and proposing a hybrid model in this field.

The importance of XAI is undeniable in the current scenario and thus it's importance in considerable. Unfortunately, there are a couple of scenarios where the current XAI models fail. Additionally, attackers seek vulnerabilities in these models. This review paper is intended to cover some of the important scenarios where the existing XAI models may fail and some progresses made which shows their vulnerabilities and how to secure them.

The application of deep learning in autonomous vehicles has surged over the years with the advances in technologies. This research explores the integration of deep learning algorithms into autonomous vehicles (AVs), focusing on their role in perception, decision-making, localization, mapping, and navigation. It shows that deep learning, which as a part of machine learning, mimics the human brain's neural networks, enabling advancements in perception, decision-making, localization, mapping, and overall navigation. Techniques like convolutional neural networks are used for image detection and steering control, while deep learning is crucial for path planning, automated parking, and traffic maneuvering. Localization and mapping are essential for AVs' navigation, with deep learning-based object detection mechanisms like Faster R-CNN and YOLO proving effective in real-time obstacle detection. Apart from the roles, the study also revealed that integration of deep learning in AVs faces challenges such as dataset uncertainty, sensor challenges, and model training intricacies. However, these issues can be addressed through increased standardization of sensors, real-life testing for model training, and advancements in model compression technologies can optimize the performance of deep learning in AVs. The study concludes that deep learning plays a crucial role in enhancing the safety and reliability of AV navigation. This study contributes to the ongoing discourse on the optimal integration of deep learning in AVs, aiming to foster their safety, reliability, and societal acceptance.

Stress has various impacts on the health of human beings. Recent success in wearable sensor development, combined with advancements in deep learning to automatically detect features from raw data, opens several interesting applications related to detecting emotional states. Being able to accurately detect stress-related emotional arousal in an acute setting can positively impact the imminent health status of humans, i.e., through avoiding dangerous locations in an urban traffic setting. This work proposes an explainable deep learning methodology for the automatic detection of stress in physiological sensor data, recorded through a non-invasive wearable sensor device, the Empatica E4 wristband. We propose an Long-Short Term-Memory (LSTM) network, extended through a Deep Generative Ensemble of conditional GANs (LSTM DGE) , to deal with the low data regime of sparsely labeled sensor measurements. As explainability is often a main concern of deep learning models, we leverage Integrated Gradients (IG) to highlight the most essential features used by the model for prediction and to compare the results to state-of-the-art expert-based stress detection methodologies in terms of precision, recall and interpretability. Results show that our LSTM DGE outperforms the state-of-the-art algorithm by 3 percentage points in terms of recall, and 7.18 percentage points in terms of precision. More importantly, through the use of Integrated Gradients as a layer of explainability, we show that there is a strong overlap between model-derived stress features for electrodermal activity, and existing literature, which current state-of-the-art stress detection systems in medical research and psychology are based on.

Training artificial intelligence (AI) for reasoning presents significant challenges due to the lack of clear objectives and reward functions. In structured environments such as chess and Go, objectives are well-defined, and the optimal moves can be determined through search algorithms that maximize the reward function. This clarity enables AI to learn and improve autonomously through random trials and recursive calculation of rewards, mirroring human learning processes that depend on prior experiences and state-dependent actions. However, in real-world applications, AI lacks access to direct interaction and feedback, making it difficult to define proper reward functions. This limitation is particularly evident in the training of large language models (LLMs), where the quality of the output is constrained by the training data and the absence of a dynamic reward system. Simulated environments offer some utility but are inherently limited by their design and scope. Achieving artificial general intelligence (AGI) requires AI to function and receive feedback in real-world settings, similar to human cognitive and strategic development through interactive experiences. This paper explores the critical role of clear objectives and reward functions in AI training, the limitations posed by the current lack of real-world interaction, and the implications for future advancements in AI reasoning capabilities.

The rapid growth of the global population is causing a significant increase in waste production, leading to serious environmental and public health challenges. To address these issues, waste management systems are incorporating advanced technologies. Machine learning and computer vision are used to predict waste patterns, optimize collection schedules, and improve sorting accuracy. Deep learning automates the sorting process, provides predictive analytics, and enhances recycling rates. Robotics, combined with AI and computer vision, improves sorting efficiency, while the Internet of Things (IoT) monitors waste levels and optimizes collection routes. Despite these benefits, challenges such as data scarcity, high computational demands, and the need for substantial infrastructure investments must be addressed. This research explores the integration of advanced technologies into waste management and evaluates their effectiveness using waste datasets. It highlights the potential to tackle environmental challenges and lay the groundwork for more intelligent waste management solutions.

This study evaluates the effectiveness of Darknet53 and VGG16 pretrained models to develop a hybrid ensemble model named HybridPlantNet23 deep learning model for effective plant disease classification, using the publicly available dataset ‘A Dataset of Leaf Images’ from Mendeley webpage. The preprocessing of the dataset involved resizing the images to the respective requirements of the pretrained models to develop a combined datastore which served as the input. All experiments were conducted using MATLAB R2023b deep network designer interface. Model performances were evaluated based on the conventional accuracy, precision, recall, and F1-score. VGG16 achieved the highest accuracy (97.25%), precision (96.30%), recall (96.96%), and F1-score (96.55%), outperforming both Darknet53 and the ensemble HybridPlantNet23 model. Despite the advanced architecture of the ensemble model, its performance (accuracy: 95.37%, precision: 93.56%, recall: 94.20%, F1-score: 93.74%) was not significantly superior, thus highlighting scientific reasons behind this development. Recommendations for future works were provided accordingly.

This study discusses the application of deep learning technology in network intrusion detection systems (IDS) and focuses on a new model named CNN-Focal. First, through the review of traditional IDS technology, it analyzes its limitations in dealing with complex network traffic. Then, the design principle of the CNN-Focal model is described in detail, which uses threshold convolution and SoftMax multi-class classification technology to effectively improve abnormal traffic detection's accuracy and efficiency. The experimental results show that CNN-Focal performs well on the open data set, demonstrating the potential and advantages of its application in the natural network environment and providing a new perspective and method for further research of deep learning in the field of network security in the future.

This study investigates anomaly detection of IoT device traffic using Convolutional Neural Networks (CNN) and Variational Autoencoders (VAE) to enhance the detection capability of security threats in IoT environments. A series of hardware configurations, software environments, and hyperparameters were utilized to optimize the training and testing processes of the models. The CNN model demonstrates robust classification performance, achieving an accuracy rate of 95.85% on the test dataset, effectively distinguishing between different types of IoT device traffic. Meanwhile, the VAE model exhibits proficient anomaly detection capabilities by effectively capturing abnormal patterns in the data using reconstruction loss and KL divergence. The combined use of CNN and VAE models offers a comprehensive solution to cybersecurity challenges in IoT environments. Future research directions include exploring diverse IoT traffic data, practical deployment for validation, and further optimization of model structures and parameters to improve performance and applicability.

This study introduces a novel method that combines Generative Adversarial Networks (GANs) with conformal prediction to revolutionize the domain of stock ticker forecasting. Our approach enhances prediction accuracy and offers dependable prediction intervals to gauge uncertainty, surpassing current GAN-based techniques in unpredictable market scenarios. We made a validation using AAPL stock data, showing the ability of our model to generate accurate forecasts and reliable assessments of prediction uncertainty. The prediction intervals generated provide investors and risk managers with a powerful tool to make informed decisions. Our approach represents a notable advancement in financial prediction as it improves the clarity and reliability of forecasts. By incorporating conformal prediction into the GAN framework, the reliability of forecasts is improved, providing a thorough evaluation of uncertainty that is crucial for successful risk management. This framework has the potential to transform the field of predictive analytics in finance. It shows a solution that balances accuracy and uncertainty quantification. The results clearly show the effectiveness of this method, paving the way for its potential application in a wide range of financial prediction scenarios.

With the widespread application of artificial intelligence technology in various industries, users' attention to privacy and data security has increased significantly. Federated learning, as a new technology paradigm combining privacy-enhanced computing and artificial intelligence, resolves the contradiction between data security and open sharing. This paper presents the benefits of federated learning in terms of privacy, real-time processing, model robustness, compliance and cross-industry applications. At the same time, when combined with Edge AI technology, federated learning promotes the decentralisation of intelligent systems, improving data privacy protection and model accuracy. This paper also discusses the application cases of federated learning in the medical field, through local data processing and model training, effectively protecting user privacy, realizing medical data sharing and model optimization, and promoting the development of artificial intelligence.

The use of Retrieval-augmented generation (RAG) using large language models (LLMs) has shown potential for addressing issues such as hallucinations and inadequately contextualized responses. A pivotal stage in the RAG process involves a retriever for retrieving chunks based on semantic similarity with the query. In this study, we advocate for and provide experimental evidence supporting integrating and maintaining questions and answers (QA) formatted databases to improve retrieved-context representations and response quality. Our experiments evaluate our approach on benchmark RAG datasets using standard evaluation metrics and provide comparative analyses against state-of-the-art retrieval methods, showing the potential of our approach.

This paper systematically discusses the application and development of federated learning in data privacy protection and data value sharing. With the rapid development of global information technology, especially the explosive growth of data from Internet of Things devices, data security and privacy protection are facing unprecedented challenges. This paper first analyzes the growth trend of global data volume and its importance to next generation technologies such as artificial intelligence technologies such as deep learning. Second, the paper provides an in-depth look at the impact of current data privacy regulations on data flows and value creation, particularly the EU's GDPR and China's Data Security and Personal Information Protection Law. Then, this paper introduces in detail federated learning, as a new distributed machine learning paradigm, which effectively solves the contradiction between existing data sharing and privacy protection by protecting individual data privacy and realizing global model collaborative construction. Finally, this paper discusses the combination of blockchain technology and federated learning, and proposes BeFL architecture as a new secure, decentralized and trusted federated learning system, which is expected to provide a comprehensive solution for large-scale data processing and value creation in multi-party scenarios. The research in this paper not only deepens the understanding of federation learning in theory, but also provides important reference and enlightenment for future research and application in related fields.

Traditional computer vision techniques aim to extract meaningful information from images but often depend on manual feature engineering, making it difficult to handle complex real-world scenarios. Fractional Calculus (FC), which extends derivatives to non-integer orders, provides a flexible way to model systems with memory effects and long-term dependencies, making it a powerful tool for capturing fractional rates of variation. Recently, Neural Networks (NNs) have demonstrated remarkable capabilities in learning complex patterns directly from raw data, automating computer vision tasks and enhancing performance. Therefore, the use of fractional calculus in neural network-based computer vision is a powerful method to address existing challenges by effectively capturing complex spatial and temporal relationships in images and videos. This paper presents a survey of Fractional Calculus Neural Network-based (FC NN-based) computer vision techniques for denoising, enhancement, object detection, segmentation, restoration, and NN compression. The survey compiles existing FFC NN-based approaches, elucidates underlying concepts, and identifies open questions and research directions. By leveraging FC's properties, FC NN-based approaches offer a novel way to improve the robustness and efficiency of computer vision systems.

Despite significant advancements in Artificial Intelligence (AI) and Large Language Models (LLMs), detecting and mitigating bias remains a critical challenge, particularly within social media platforms like X (formerly Twitter) and in addressing cyberbullying present on them. This research investigates the effectiveness of leading LLMs in generating synthetic biased and cyberbullying data and evaluates the proficiency of Transformer AI models in detecting bias and cyberbullying within both authentic and synthetic contexts. The study involves semantic analysis and feature engineering on a dataset of over 48,000 sentences related to cyberbullying collected from Twitter (before it became X). Leveraging state-of-the-art LLMs such as ChatGPT-4o, Pi AI, Claude 3 Opus, and Gemini-1.5, synthetic biased, cyberbullying, and neutral data were generated to deepen the understanding of bias in human-generated data. AI models including DeBERTa, Longformer, BigBird, HateBERT, MobileBERT, DistilBERT, BERT, RoBERTa, ELECTRA, and XLNet were initially trained to classify Twitter cyberbullying data and subsequently fine-tuned, optimized, and quantized for multilabel classification (detecting both biases and cyberbullying). The study's outcomes include a prototype of a hybrid application that combines a Bias Data Detector and a Bias Data Generator, validated through extensive testing.

This paper presents a comprehensive comparative analysis of four different computational vision algorithms for implementing an unpalleting system in a factory environment, under typical edge computing constraints. The primary objective is to automate the process of detecting and locating specific variable-shaped objects on a pallet, enabling a robotic system to accurately unstack them. The four algorithms evaluated are Pattern Matching, Scale-Invariant Feature Transform, Oriented FAST and Rotated BRIEF, and Haar cascade classifiers. Each technique is described in detail in this paper, and their implementations are outlined. Experimental results are thoroughly analyzed, assessing the algorithms’ performance based on accuracy, robustness to variability, computational speed, detection sensitivity, and resource consumption. The findings reveal the strengths and limitations of each algorithm, providing valuable insights for selecting the most appropriate technique based on the specific requirements of the unpalleting system. This research contributes to the advancement of automated object detection and localization in industrial environments, paving the way for more efficient and reliable unpalleting systems.

This paper presents two novel bio-inspired particle swarm optimisation (PSO) variants: biased eavesdropping PSO (BEPSO) and altruistic heterogeneous PSO (AHPSO). These algorithms are inspired by types of group behaviour found in nature that have not previously been exploited in search algorithms. The primary search behaviour of the BEPSO algorithm is inspired by eavesdropping behaviour observed in nature coupled with a cognitive bias mechanism that enables particles to make decisions on cooperation. The second algorithm, AHPSO, conceptualises particles in the swarm as energy-driven agents with bio-inspired altruistic behaviour which allows the formation of lending-borrowing relationships. The mechanisms underlying these algorithms provide new approaches to maintaining swarm diversity which contributes to preventing premature convergence. The new algorithms were tested on the 30, 50 and 100-dimensional CEC'13, CEC'14 and CEC'17 test suites, and various constrained real-world optimisation problems, against 13 well-known PSO variants and the CEC competition winner, differential evolution algorithm L-SHADE. The experimental results show that both algorithms, BEPSO and AHPSO, provide very competitive performance on the unconstrained test suites and the constrained real-world problems. They were significantly better than most other PSO variant on most problem sets and no other comparator algorithm was significantly better than either of them on any of the 50 and 100-d problem sets.

The article evaluates the use of machine learning methods to predict valve plate failures in hydraulic pumps. To conduct this assessment, a laboratory setup was developed that allowed the creation of a training dataset containing both the normal operating state of the pump and the operating state with three different valve plate damages. Five different predictive models were evaluated, including three composite models based on decision trees: Random Forest, Gradient-Boosted Trees, and Rotation Forest, as well as the kNN algorithm and a neural network. The analysis showed that neural networks achieved the highest prediction accuracy, enabling a prediction accuracy level of 89%. The kNN algorithm ranked second, and tree-based algorithms performed 4% worse than the neural network. Additionally, the attribute importance analysis revealed that leak flow, pressure output together with output flow, are the most important parameters for accurate prediction.

Accurate prostate volume estimation is crucial for effective prostate disease management. Ultrasound (US) imaging, particularly transrectal ultrasound, offers a cost-effective and rapid assessment. However, US images often suffer from artifacts and poor contrast, making prostate volume estimation challenging. This review explores recent advancements in deep learning (DL) techniques for automatic prostate segmentation in US images as a primary step for prostate volume estimation. We examine various DL architectures, including traditional U-Net modifications and innovative designs incorporating residual connections, multi-directional image data, and attention mechanisms. Additionally, we discuss pre-processing methods to enhance image quality, the integration of shape information, and strategies to improve the consistency and robustness of DL models. The effectiveness of these techniques is evaluated through metrics such as Dice Similarity Coefficient, Jaccard Index, and Hausdorff Distance. The review highlights the potential of DL in improving prostate volume estimation accuracy and reducing clinical workload, while also identifying areas for future research to enhance model performance and generalizability.

With the advance of generative AI, the text-to-image (T2I) model has the ability to generate various contents. However, the generated contents cannot be fully controlled. There is a potential risk that text-to-image model can generate unsafe images with uncomfortable contents. In our work, we focus on eliminating the NSFW (not safe for work) content generation from text-to-image model while maintaining the high quality of generated images by fine-tuning the pretrained diffusion model via reinforcement learning by optimizing the well-designed content-safe reward function. The proposed method leverages a customized reward function consisting of the CLIP (Contrastive Language-Image Pre-training) and nudity rewards to prune the nudity contents that adhere to the pretrained model and keep the corresponding semantic meaning on the safe side. In this way, the text-to-image model is robust to unsafe adversarial prompts since unsafe visual representations are mitigated from latent space. Extensive experiments conducted on different datasets demonstrate the effectiveness of the proposed method in alleviating unsafe content generation while preserving the high-fidelity of benign images as well as images generated by unsafe prompts. Our method surpasses existing state-of-the-art (SOTA) baseline methods and achieves better performance on sexual content removal (14.5% higher than SafeGen) and image quality retention. In terms of robustness, our method outperforms Safe Latent Diffusion and SafeGen under the SOTA black-box attacking model SneakyPrompt by approximately 2x and 5.6x, respectively. Furthermore, our constructed method can be a benchmark for other methods aiming for anti-NSFW generation, high-level prompt-image safe alignment.

Artificial intelligence provides a new research concept for digital image processing. However, at present, artificial intelligence is rarely introduced into the teaching of digital image processing in colleges and universities, and there are problems such as obsolete teaching content, single teaching methods, and simple course experiments, which affect the teaching effect and are not conducive to the cultivation of comprehensive and innovative talents. Digital image processing technology brings more possibilities to communication engineering and makes communication more convenient. For example, video calls and photo transmission make people's communication methods in daily life more and more diversified. Time and space limitations allow people to meet online, creating more communication possibilities. However, there are still many problems and methods worthy of in-depth exploration. Therefore, this paper has a comprehensive understanding and mastery of the traditional and deep learning methods of digital image processing to improve the relevant project practice and scientific research exploration ability and refer for similar research conclusions.

We study how abstract representations emerge in a Deep Belief Network (DBN) trained on benchmark datasets. Our analysis targets the principles of learning in the early stages of information processing, starting from the “primordial soup" of the under-sampling regime. As the data is processed by deeper and deeper layers, features are detected and removed, transferring more and more “context-invariant” information to deeper layers. We show that the representation approaches an universal model – the Hierarchical Feature Model (HFM) – determined by the principle of maximal relevance. Relevance quantifies the uncertainty on the model of the data, thus suggesting that “meaning" – i.e. syntactic information – is that part of the data which is not yet captured by a model. Our analysis shows that shallow layers are well described by pairwise Ising models, which provide a representation of the data in terms of generic, low order features. We also show that plasticity increases with depth, in a similar way as it does in the brain. These findings suggest that DBNs are capable of extracting a hierarchy of features from the data which is consistent with the principle of maximal relevance.

Aiming at the problems of inaccurate segmentation of long object and information loss of small object in real-time semantic segmentation algorithm, this paper proposes a lightweight multi-branch real-time semantic segmentation network based on BiseNetV2. The new auxiliary branch makes full use of spatial details and context information to cover the long object in the field of view. Meanwhile, in order to ensure the inference speed of the model, the asymmetric convolution is used in each stage of the auxiliary branch to design a structure with low computational complexity. In the multi-branch fusion stage, the alignment and fusion module is designed to provide guidance information for deep and shallow feature mapping, so as to make up for the problem of feature misalignment in the fusion of information at different scales, and thus reduce the loss of small target information. In order to further improve the model’s awareness of key information, a global context module is designed to capture the most important features in the input data. The proposed network uses NVIDIA GeForce RTX 3080 Laptop GPU experiment, on the road street view data set Cityscapes and CamVid average occurring simultaneously ratio reached 77.1% and 78.4% respectively, with running speed of 127 frames/s respectively and 112 frames/s. The experimental results show that the proposed algorithm can achieve real-time segmentation and improve the accuracy significantly, showing good semantic segmentation performance

RNA sequencing techniques, like bulk RNA-seq and Single Cell (sc) RNA-seq, are critical tools for the biologist looking to analyze the genetic activity/transcriptome of a tissue or cell during an experimental procedure. Platforms like Illumina's next-generation sequencing (NGS) are used to produce the raw data for this experimental procedure. This raw FASTQ data must then be prepared via a complex series of data manipulations by bioinformaticians. This process currently takes place on an unwieldy textual user interface like a terminal/command line that requires the user to install and import multiple program packages, preventing the untrained biologist from initiating data analysis. Open-source platforms like Galaxy have produced a more user-friendly pipeline, yet the visual interface remains cluttered and highly technical, remaining uninviting for the natural scientist. To address this, SeqMate is a user-friendly tool that allows for one-click analytics by utilizing the power of a large language model (LLM) to automate both data preparation and analysis (differential expression, trajectory analysis, etc). Furthermore, by utilizing the power of generative AI, SeqMate is also capable of analyzing such findings and producing written reports of upregulated/downregulated/user-prompted genes with sources cited from known repositories like PubMed, PDB, and Uniprot.

The traffic light detection and recognition are crucial for enhancing the security of unmanned systems. This study proposes a YOLOv5-based traffic light detection algorithm to tackle the challenges posed by small targets and complex urban backgrounds. Initially, the mosaic-9 method is employed to enhance the training dataset, thereby boosting the network’s ability to generalize and adapt to real-world scenarios. Furthermore, the network incorporates the Squeezed-and-Excitation (SE) attention mechanism to improve. Moreover, the YOLOv5 algorithm’s loss function has been optimized by substituting it with EIoU_loss, which addresses issues like missed detection and false alarms. Experimental results demonstrate that the model , trained with this enhanced network, achieves a 99.4% mAP on a custom dataset, which is 6.3% higher than the original YOLOv5, while maintaining a detection speed of 74 f/s. Therefore, this algorithm offers higher detection accuracy and effectively meets real-time operational requirements.

We present a novel benchmarking methodology for Large Language Models (LLMs) to evaluate their susceptibility to hallucinations, thereby determining their reliability for real-world applications involving greater responsibilities. This method, called Deception-Based Benchmarking, involves testing the model with a task that requires composing a short paragraph. Initially, the model performs under standard conditions. Then, it is required to begin with a misleading sentence. Based on these outputs, the model is assessed on three criteria: accuracy, susceptibility, and consistency. This approach can be integrated with existing benchmarks or applied to new ones, thus facilitating a comprehensive evaluation of models across multiple dimensions. It also encompasses various forms of hallucination. We applied this methodology to several small opensource models using a modified version of MMLU, DB-MMLU1 . Our findings indicate that most current models are not specifically designed to self-correct when the random sampling process leads them to produce inaccuracies. However, certain models, such as Solar-10.7B-Instruct, exhibit a reduced vulnerability to hallucination, as reflected by their susceptibility and consistency scores. These metrics are distinct from traditional benchmark scores. Our results align with TruthfulQA, a widely used benchmark for hallucination. Looking forward, DB-benchmarking can be readily applied to other benchmarks to monitor the advancement of LLMs.

This paper reviews the multiple roles of machine learning in improving the resilience of power grids, especially in applying new energy storage technologies. Energy storage technologies, such as compressed air energy storage, flywheel energy storage, and superconducting coil energy storage, significantly improve the power grid's ability to respond to load fluctuations and emergencies through intelligent control and optimisation of machine learning algorithms. This not only helps to optimise energy dispatch and improve the operational efficiency and flexibility of the grid but also significantly improves the stability and reliability of the grid so that it is better able to meet future challenges and needs.

In this paper, we present a novel approach for text independent phone-to-audio alignment based on phoneme recognition, representation learning and knowledge transfer. Our method leverages a self-supervised model (wav2vec2) fine-tuned for phoneme recognition using a Connectionist Temporal Classification (CTC) loss, a dimension reduction model and a frame-level phoneme classifier trained thanks to forced-alignment labels (using Montreal Forced Aligner) to produce multi-lingual phonetic representations, thus requiring minimal additional training. We evaluate our model using synthetic native data from the TIMIT dataset and the SCRIBE dataset for American and British English, respectively. Our proposed model outperforms the state-of-the-art (charsiu) in statistical metrics and has applications in language learning and speech processing systems. We leave experiments on other languages for future work but the design of the system makes it easily adaptable to other languages.

Quantum machine learning applications have become viable with the recent advancements in quantum computing. Merging ML with the power of quantum computing holds great potential for data-driven decision-making, as well as the development of more powerful models capable of handling more complex datasets with faster processing time. This area offers potential for improving the accuracy of real-time forecasting renewable energy production. However, the literature on this topic is sparse. Addressing this knowledge gap, this study aims to design, implement, and evaluate the performance of a quantum neural network forecast model for solar irradiance up to 3-hours ahead. The proposed model was compared with Support Vector Regression, Group Method of Data Handling, and Extreme Gradient Boost classical models. Using the best quantum neural network configuration found, the proposed framework could provide competitive results when compared to its competitors, considering forecasting intervals of 5- to 120-minutes ahead, where it was the fourth best-performing paradigm. For 3-hours ahead predictions, the QNN was able to overcome its clas-sical counterparts, but the XGBoost. This fact can be an indication that the quantum model may identify and retrieve relevant spatiotemporal information from the input dataset in such a manner not attainable by the current classical approaches.

A significant challenge encountered in mmWave and Sub-terahertz systems used in 5G and 1 the upcoming 6G networks is the rapid fluctuation in signal quality across various beam directions. Extremely high-frequency waves are highly vulnerable to obstruction, making even slight adjustments in device orientation or the presence of blockers capable of causing substantial fluctuations in link quality along a designated path. This issue poses a major obstacle because numerous applications with low-latency requirements necessitate precise forecasting of network quality from many directions and cells. The method that is proposed in this research demonstrates an avant-garde approach for assessing the quality of multi-directional connections in mmWave systems by utilizing the Liquid Time Constant network (LTC) instead of the conventionally used Long Short-Term Memory (LSTM) technique. The method’s validity was tested through an optimistic simulation involving the monitoring of multi-cell connections at 28 GHz in a scenario where humans and various obstructions were moving arbitrarily. The results with LTC are significantly better than those obtained by conventional approaches such as LSTM. The latter resulted in a test Root Mean Squared Error (RMSE) of 3.44 dB, while the former, 0.25 dB, demonstrating a 13-fold improvement. For better interpretability and to illustrate the complexity of prediction, an approximate mathematical expression is also fitted to the simulated signal data using Symbolic Regression.

Artificial intelligence technologies based on deep convolutional neural networks and large lan-guage models have made significant breakthroughs in many tasks such as image recognition, target detection, semantic segmentation and natural language processing, but also face the problem of contradiction between the high computational capacity of the algorithms and the limited deployment resources. Quantization, which converts floating-point neural networks into low-bit-width integer networks, is an important and essential technique for efficient deploy-ment and cost reduction in edge computing. This paper analyze various existing quantization methods, showcase the deployment accuracy of advanced techniques, and discuss the future challenges and trends in this domain.

The paper explores using generative artificial intelligence (AI) in financial market data management and forecasting. By integrating multiple data sources and feature extraction techniques, such as fundamental analysis, technical indicators, global economic data, and sentiment analysis, generative AI constructs a comprehensive deep learning framework that significantly enhances financial data management efficiency and market forecasts' accuracy. Specifically, technologies like generative adversarial networks (Gans) and variational autoencoders (VAE) demonstrate substantial data augmentation and model optimisation potential. The application value of the model in real-time market prediction and trading strategy optimization is further amplified through reinforcement learning methods.

This article examines the integration of artificial intelligence (AI) and robotic process automation (RPA) in financial accounting and management, underscoring their role in driving the digital transformation of corporate finance. It discusses the shortcomings of traditional financial processes and highlights the potential of AI and RPA technologies to enhance efficiency, accuracy, and cost-effectiveness. The paper also explores the limitations of RPA, such as its challenges in processing unstructured data and handling complex decision-making scenarios. Looking forward, it considers the future trends in AI and RPA, emphasizing the benefits of cloud technology in scaling automated systems and addressing associated challenges.

This paper presents a comprehensive study on applying machine learning (ML) techniques for real-time credit card fraud detection. With the increasing prevalence of credit card fraud, financial institutions face the challenge of detecting fraudulent transactions promptly and accurately. This research evaluates various ML models, including Logistic Regression, Decision Trees, Random Forests, XGBoost, and Deep Convolutional Neural Networks, for their effectiveness in identifying fraudulent activities. Utilizing PySpark for processing large-scale transaction data, the study highlights the importance of real-time analysis, adaptive learning, and handling imbalanced datasets. Key findings reveal that XgBoost, with its balance of accuracy and complexity, emerged as the most promising model, achieving an accuracy rate of 98% in detecting fraudulent transactions. The paper further discusses the potential of ensemble methods, graph-powered systems, and intelligent sampling in enhancing fraud detection capabilities. Challenges such as overfitting, data access, and the need for real-time analysis are addressed. Future research directions include implementing models in live transaction environments, expanding model applicability across financial platforms, and developing advanced anomaly detection methodologies. This study underscores the pivotal role of machine learning in safeguarding financial transactions against fraud, offering significant implications for consumers, financial institutions, and the broader financial ecosystem.

This review provides an overview of eXplainable AI (XAI) methods for oncological ultrasound image analysis and compares their performance evaluations. A systematic search of Medline Embase and Scopus between March 25 and April 14 2024 identified 17 studies describing 14 XAI methods, including visualization, semantics, example-based, and hybrid functions. These methods primarily provided specific, local, and post-hoc explanations. Performance evaluations focused on AI model performance, with limited assessment of explainability impact. Standardized evaluations incorporating clinical end-users are generally lacking. Enhanced XAI transparency may facilitate AI integration into clinical workflows. Future research should develop real-time methodologies and standardized quantitative evaluative metrics.

This paper presents a comparative study of fundamental similarity functions for Siamese networks in semantic textual similarity (STS) tasks. We evaluate various similarity functions using the STS Benchmark dataset, analyzing their performance and stability. Additionally, we introduce a multi-objective approach for optimal threshold selection. Our findings provide insights into the effectiveness of different similarity functions and offer a straightforward method for threshold selection optimization, contributing to the advancement of Siamese network architectures in STS applications.

The focus was on developing machine learning models to predict the IC50 values of the compound OTX015 using gene expression levels of cancer cell lines. Through regression-based machine learning models, particularly the SVM model, trained on pre-laboratory data, consistent and generalizable performance with low error scores was achieved. These models underwent enhancement via hyperparameter optimization and were tested on validation data. The results indicate the utility of machine learning in drug discovery processes and personalized medicine, offering rapid and cost-effective predictions of IC50 values. However, for enhanced reliability, further model development, data gathering, and evaluation with diverse compounds are recommended. In conclusion, this research demonstrates the potential of machine learning approaches to optimize experimental processes, reduce costs, and contribute to personalized treatments in healthcare by accurately predicting compound responses.

The integration of deep learning models into financial risk prediction and analysis has significantly transformed traditional approaches. While conventional quantitative methods often rely on simplistic metrics like maximum drawdown, the advent of deep learning necessitates a more nuanced evaluation, emphasizing the model's generalization ability, especially during market crises such as stock market crashes. This paper explores the critical aspects of evaluating deep learning models' risk control capabilities in finance, underscoring the importance of understanding both statistical metrics and generalization abilities, particularly in adverse market conditions. By examining deep learning models' performance in scenarios like stock market crashes and highlighting the significance of cross-validation techniques, this study aims to offer practitioners insights into constructing robust risk management systems. It advocates for a comprehensive approach integrating quantitative analysis with macroeconomic factors to enhance financial risk prediction and analysis in volatile markets. The experimentation reveals that different deep generative models excel in various aspects of financial time series analysis, with generative adversarial networks (GANs) demonstrating superior performance in predicting Value at Risk (VaR) and variational autoencoders (VAEs) excelling in return rate prediction. Moreover, integrating multiple models further enhances predictive performance, leveraging the strengths of each model to compensate for individual weaknesses. Overall, this paper underscores the potential and significance of deep generative models in financial time series analysis, offering a roadmap for improved risk management and decision-making in financial markets.

Securing funding is a critical yet complex task for organizations and individuals. This study presents an innovative chatbot designed to streamline the process using advanced Natural Language Processing (NLP) techniques and, specifically, a Retrieval-Augmented Generation (RAG) pipeline optimized for real-world applications. Our chatbot assists users in identifying suitable public tenders for financial support through natural language queries and a comprehensive public data database. The chatbot operates in a two-stage interaction model, initially providing summarized tender information for exploratory brainstorming, followed by detailed data upon user selection. A custom filtering mechanism ensures that the user interface elements responsible for swapping the interaction stages are consistent with the responses generated by the conversational agent. Human-evaluation tests demonstrated an average accuracy of 90.4% in document retrieval, with an average of 2.11 interactions required to find a specific tender. User satisfaction, rated on a scale of 1 to 5, averaged 3.14 (±1.73), indicating generally positive user experience with room for improvement. This approach addresses challenges of relevance, accuracy, and conversational flow, resulting in a reliable chatbot that simplifies the process of finding funding opportunities.

YOLOv8, as an efficient object detection method, can swiftly and precisely identify objects within images. However, traditional algorithms encounter difficulties when detecting small targets in remote sensing images, such as missing information, background noise, and interactions among multiple objects in complex scenes, which may affect performance. To tackle these challenges, we propose an enhanced algorithm optimized for detecting small objects in remote sensing images, named HP-YOLOv8. Firstly, we design the C2f-D-Mixer (C2f-DM) module as a replacement for the original C2f module. This module integrates both local and global information, significantly improving the ability to detect features of small objects. Secondly, we introduce a feature fusion technique based on attention mechanisms, named Bi-Level Routing Attention in Gated Feature Pyramid Network (BGFPN). This technique utilizes an efficient feature aggregation network and reparameterization technology to optimize information interaction between different scale feature maps, and through the Bi-level Routing Attention (BRA) mechanism, it effectively captures critical feature information of small target objects. Finally, we propose the Smooth Mean Perpendicular Distance Intersection over Union (SMPDIoU) loss function. The method comprehensively considers the shape and size of detection boxes, enhances the model's focus on the attributes of detection boxes and provides a more accurate bounding box regression loss calculation method. To demonstrate our approach's efficacy, we conduct comprehensive experiments across the RSOD, NWPU VHR-10, and VisDrone2019 datasets. The experimental results show that the HP-YOLOv8 achieves 95.11%, 93.05%, and 53.49% in the [email protected] metric, and 72.03%, 65.37%, and 38.91% in the more stringent [email protected]:0.95 metric, respectively.

Cryptocurrencies have now become an emerging blockchain-based payment technology. Users’ identities on such networks are pseudo-anonymous in that all transactions made from an address are transparent and searchable by anyone; to preserve their privacy, users often use many different addresses. In recent years, some studies have been conducted regarding analyzing clusters of Bitcoin addresses that, according to certain heuristics, belong to the same entity. Such action allows law enforcement to have relevant information in cases where cryptocurrencies are used for illegal trafficking. Clustering methods based on a single heuristic do not allow the clustering of many addresses in a complete and accurate manner. This paper proposes Bitcoin Address Clustering based on multiple Heuristics (BACH), a tool that uses three different clustering heuristics to identify clusters of Bitcoin addresses, which are displayed through a three-dimensional graph. Based on the results, several analyses were conducted, including a comparison with Wallet Explorer, an address clustering tool similar to BACH. BACH, in addition to introducing the innovative feature of graphical visualization of the internal structure of clusters, is shown to improve address aggregation through the joint use of three heuristics for clustering.

Abstract — Game development using technology was started in 1950 around the first video game created Tennis for Two by Higginbotham as a fun diversion, to show the power of technology. The advanced development of Artificial intelligence (AI), has turned out to be the heart of game development technology and AI has revolutionized gaming and brought so many features to life.in reality, it isn't easy to give a definition and meaning to it even though games are becoming a part of human survival. It is difficult to define the real meaning of a game but it is part of human survival as it usually refers to the cleverness of the way the human being involvement in the characters of the game. As a part of game development, many technologies were used to create a game for the players with creativity, complex design, variety of levels, and aspects for the player to use choices effectively and efficiently to interact with the character in the game and enhance the experience to the next levels. This paper initially gives the importance of game development and need for the technology usage to enhance the current state of games. Then analyzes the history and present state of Artificial intelligence in game development. Finally, it explains the highlights of potential changes in game development and the impact and influence of artificial intelligence and machine learning models in the future development of games.

Abstract: Online services have transformed into a boon for various industries and globalization as a whole. One of the ventures reliant upon the Web is finance. Pattern prediction and risk reduction in the stock market and other financial decisions are made possible by some technologies. These technologies promise to expand innovation, knowledge, and efficiency opportunities in the financial sector. Finance frequently uses machine learning technology to identify risks based on probabilistic statistics and historical data to support investment decisions. Additionally, it can be utilized for risk assessment and risk management planning. Internet technology, which enables real-time data analysis, streamlined transactions, and improved customer communication, has particularly benefited the financial sector. However, these advancements come with significant challenges, such as the need for robust online credit protection and security threats. This study looks at how advanced machine learning methods are used in the financial sector and focuses on how they have changed credit scoring, algorithmic trading, and predictive modeling.

This article continues research efforts on performance and energy efficiency of intelligent algorithms and its role for green and sustainable computing. The introduction focusses on Bremermann's Limit and its effect on extensive approach for improvement of computers performance. The article aims to identify role of Intelligent Algorithms energy efficiency, how it differs from general software energy efficiency. An improved empirical investigation on heuristic methods for search and optimisation illustrates algorithms' energy efficiency. Experimental results and consideration of further work conclude the article.

With the continuous advancement of digital transformation, digital payments are playing an increasingly important role in the financial industry. This study aims to utilize machine learning models to predict and analyze digital payment behavior. Initially, the background and significance of digital payments in the financial sector are introduced. Subsequently, the current status and trends of traditional digital payment distribution are reviewed, alongside related work on digital payment behavior prediction. Methodologically, principles and applications of machine learning models such as logistic regression, decision trees, and random forests are elaborated, along with experimental design and data preprocessing methods. The experimental results and discussion section illustrates the performance of each model in digital payment prediction and explores their impact on credit decisions. This exploration equips financial institutions with more effective user behavior analysis and risk management tools, thereby fostering future development and application of digital payment technologies.

We developed a novel machine learning (ML) algorithm with the goal of producing transparent models (i.e. understandable-by-humans) while also flexibly accounting for nonlinearity and interactions. Our method is based on ranked sparsity, and allows for flexibility and user-control in varying the shade of the opacity of black-box machine learning methods. The main tenet of ranked sparsity is that an algorithm should be more skeptical of higher-order polynomials and interactions \textit{a~priori} compared to main effects, and hence the inclusion of these more complex terms should require a higher level of evidence. In this work, we put our new ranked sparsity algorithm (as implemented in the open-source R package, `sparseR`) to the test in a predictive model "bakeoff" (i.e. a benchmarking study of ML algorithms applied "out-of-the-box," that is, with no special tuning). Algorithms were trained on a diverse set of simulated and real-world data sets from the Penn Machine Learning Benchmarks database, addressing both regression and binary classification problems. We evaluate the extent to which our human-centered algorithm can attain predictive accuracy that rivals popular black-box approaches such as neural networks, random forests, and support vector machines, while also producing more interpretable models. Using out-of-bag error as a meta-outcome, we describe the properties of data sets in which human-centered approaches can perform as well as or better than black-box approaches. We find that interpretable approaches predicted optimally or within 5% of the optimal method in most real-world data sets. We provide a more in-depth comparison of the performances of random forests to interpretable methods for several case studies, including exemplars in which algorithms performed similarly, and several cases when interpretable methods underperformed. This work provides a strong rationale for including human-centered transparent algorithms such as ours in predictive modeling applications.

To enhance the accuracy and efficiency of address parsing in traffic accident handling, this study designs a deep address parsing model based on BERT, named ASSPM. This model effectively addresses the limitations of existing deep learning models in Chinese address parsing tasks by integrating address spatial semantics. First, a lexicon-enhancement method based on hybrid representation is employed, transforming the spatial semantics extracted by the Address Spatial Semantics Extractor (ASSE) into a symbolic semantic space that can be integrated with the feature vector space of the deep model. This achieves effective fusion of spatial semantics and the deep model at the model level. Second, a lexicon adapter enhanced with spatial semantics is utilized to integrate spatial semantics into the Transformer layers of the BERT model, achieving semantic fusion at the BERT level. Finally, comparative experiments on a custom address dataset and a public Chinese address dataset validate the proposed method’s superior performance over other baseline methods. This research provides robust support for the application of deep address parsing models integrating address spatial semantics in traffic accident handling, promising to improve the efficiency and accuracy of traffic accident management.

Abstract: Hand gestures are an essential kind of nonverbal communication that let people engage and be expressive with one another. As technology has progressed, hand gesture recognition systems have become an important field of study, providing a link between human activities and artificial intelligence. An overview of the field of hand gesture recognition systems is given in this abstract, along with an explanation of its importance, difficulties, and possible uses. This abstract begins by examining the role that hand gestures have in communication. It then goes on to show the cultural significance and wide range of variations in gestural expressions that occur in many contexts and societies. Artificial intelligence systems face a unique problem in comprehending and interpreting these motions, which calls for the combination of computer vision, machine learning, and pattern recognition approaches. The abstract goes on to address why people develop hand motions. The purpose behind creating systems to recognize hand gestures is also covered in the abstract, with a focus on how these systems could improve human-computer interaction in several fields, including robotics, virtual reality, healthcare, and assistive technologies. These systems have the potential to enhance accessibility, efficiency, and user experience by allowing people to engage with devices and interfaces naturally and intuitively. The abstract also discusses environmental elements including backdrop clutter and lighting, as well as the technological difficulties in identifying and understanding hand gestures, which include variations in hand forms, orientations, and movements. To tackle these obstacles, strong algorithms that can record and interpret complex hand movements in real-time must be developed.

Determining the general-name-derived semantic relationships between geographic names often requires consulting many geographic name documents for verification. In geographic name translation and geographic name knowledge graphs, this method cannot meet the needs of large-scale extraction of general-name-derived semantic relationships between geographic entities. To solve this problem, this paper proposes a general-name derivation relation recognition method based on Prompt learning and Semantic Spatiotemporal Distribution Attention Feature Fusion (SSD-AFF). This method adopts a pipe-based relationship recognition strategy. First, the Prompt learning method is adopted to form a general-name derivation pattern with the general-name-derived geographical name category and the original geographical name category as the prompt word, and then the derivation pattern is combined with the geographical name as the input of the model so that the recognition problem of the general-name derivation of the geographical name can be transformed into a classification problem using the bidirectional language model. Then, by analyzing the characteristics of the semantic association and geographical spatio-temporal distribution between the general-name-derived geographical name and the original geographical name, the semantic spatio-temporal distribution characteristics of the general-name-derived geographical name are constructed. Finally, the network model based on the semantic spatio-temporal distribution Attention Feature Fusion (SSD-AFF) is used to distinguish the general-name-derived semantic relationships. The experimental results show that the proposed method has good performance. It can effectively mine the general-name-derived semantic relations between the vector data of geographical entities and has important application value in geographical name translation and geographical name knowledge maps.

Natural language processing model (NLP) are used in chatbots to understand user input, interpret its meaning, and generate conversational responses to provide immediate and consistent assistance. This reduces problem-solving time and staff workload and increases user satisfaction. There are both rule-based chatbots, which use decision trees and are programmed to answer specific questions, and self-learning chatbots, which can handle more complex conversations through continuous learning about data and user interactions. However, only a few chatbots have been developed specifically for the Italian language. This work proposes an NLP model to develop a powerful and efficient Italian QA (Question Answering) chatbot that is easy to use for Italian Public Administration (PA). The proposed model is based on BERT (Bidirectional Encoder Representations from Transformer) architecture, where an Encoder/Decoder module and a Highway Network module have been added to perform more efficient filtering of input text. The Italian version of the Stanford Question Answering Dataset (SQuAD-IT) is used to test the proposed model. The proposed model is one of the first models developed for Italian language-specific chatbots.

: Multimodal Large Language Models (MLLMs) harness comprehensive knowledge spanning text, images, and audio to adeptly tackle complex problems, including zero-shot in-context learning scenarios. This study explores the ability of MLLMs in visually solving the Traveling Salesman Problem (TSP) and Multiple Traveling Salesman Problem (mTSP) using images that portray point distributions on a two-dimensional plane. We introduce a novel approach employing multiple specialized agents within the MLLM framework, each dedicated to optimizing solutions for these combinatorial challenges. Our experimental investigation includes rigorous evaluations across zero-shot settings and introduces innovative multi-agent zero-shot in-context scenarios. The re-sults demonstrated that both multi-agent models—Multi-Agent 1, which includes the Initializer, Critic, and Scorer agents, and Multi-Agent 2, which comprises only the Initializer and Critic agents—significantly improved solution quality for TSP and mTSP problems. Multi-Agent 1 ex-celled in environments requiring detailed route refinement and evaluation, providing a robust framework for sophisticated optimizations. In contrast, Multi-Agent 2, focusing on iterative re-finements by the Initializer and Critic, proved effective for rapid decision-making scenarios. These experiments yield promising outcomes, showcasing the robust visual reasoning capabilities of MLLMs in addressing diverse combinatorial problems. The findings underscore the potential of MLLMs as powerful tools in computational optimization, offering insights that could inspire further advancements in this promising field.

This article continues research efforts on performance and energy efficiency of intelligent algorithms and its role for green and sustainable computing. The introduction focusses on Bremermann's Limit and its effect on extensive approach for improvement of computers performance. The article aims to identify role of Intelligent Algorithms energy efficiency, how it differs from general software energy efficiency. An improved empirical investigation on heuristic methods for search and optimisation illustrates algorithms' energy efficiency. Experimental results and consideration of further work conclude the article.

In Southeast Asia, the incidence of Leukemia, a malignant blood cancer originating from hema-topoietic progenitor cells, is on the rise, marked by a concerning 54% mortality rate. This study focuses on enhancing early-stage prediction to improve patient recovery prospects significantly. Leveraging Machine Learning and Data Science, we employ protein sequential data from frequently mutated genes such as BCL2, HSP90, PARP, and RB to predict Chronic Myeloid Leukemia (CML). Our approach relies on robust feature extraction techniques, namely Di-peptide Composition (DPC), Amino Acid Composition (AAC), and Pseudo amino acid composition (Pse-AAC), with prior attention to addressing outliers and validating feature selection through the Pearson Corre-lation Coefficient. Data augmentation ensures a well-rounded dataset for analysis. Employing a range of Machine Learning models, including Support Vector Machine (SVM), XGBoost, Random Forest (RF), K Nearest Neighbor (KNN), Decision Tree (DT), and Logistic Regression (LR), we achieve accuracy rates spanning from 66% to 94%. These classifiers undergo comprehensive as-sessment using performance metrics such as accuracy, sensitivity, specificity, F1-score, and the confusion matrix. Our proposed solution, encompassing a user-friendly web application dashboard, presents an invaluable tool for early CML diagnosis with profound implications for practitioners, offering a deploy-able asset within healthcare institutions and hospitals.

This article explores the application of machine learning techniques, specifically focusing on ensemble methods like Random Forests, for detecting fraudulent activities in digital financial transactions. Highlighting the evolution from traditional statistical approaches to modern machine learning models, it underscores the effectiveness of Random Forests in handling the inherent challenges of imbalanced datasets typical in fraud detection scenarios. Using a Kaggle dataset of credit card transactions, the study optimizes Random Forest parameters through rigorous parameter tuning, achieving significant improvements in model performance metrics such as Area Under the Curve (AUC). The findings underscore the critical role of machine learning in enhancing fraud detection capabilities, emphasizing the ongoing evolution and future potential of these methodologies in financial risk management.

This document introduces methods and algorithms to characterize a series of values as patterns of repeating symbols or sequences selected with the criterion to maximize the information retrieved. The method converts a series of real-number values into texts. Then, it interprets the text from a point of view specified by several parameters. Located patterns are linked nested, with shorter sequences within longer ones. Texts are then synthesized and organized in a tree-like structure, substantially reducing entropy. The characterization of processes serves as the basis for constructing models to estimate likely projected values. Alternative ways to assess the multiscale complexity of unidimensional processes are applications of the results obtained here.

This paper presents MADS (Multi-Agents for Data Science), an innovative and simple open-source multi-agent framework designed for systemic pipeline execution in applied supervised machine learning. By leveraging the capabilities of multi-agent systems (MAS), we introduce a universal approach to optimize and streamline machine learning pipelines. Our framework highlights the differences between various types of agents, such as reinforcement learning (RL) agents and large language model (LLM) agents, and their distinct contributions to the process. While we currently employ LLM agents to automate and enhance machine learning tasks, we acknowledge the potential of incorporating RL agents in future iterations to further improve performance and adaptability. The primary objective is to enhance the efficiency, scalability, and adaptability of supervised learning applications across various domains. This integration addresses the complexity and manual effort typically associated with machine learning workflows, paving the way for more automated, robust, and scalable solutions. Our approach demonstrates significant improvements in task automation, reduced human intervention, and enhanced model performance. The MADS framework, which will soon be available as an open-source implementation, represents a pivotal contribution to the field of machine learning, promising to facilitate broader adoption and collaborative advancement.

As large language models (LLMs) continue to advance, evaluating their comprehensive capabilities becomes significant for their application in various fields. This research study comprehensively evaluates the language, vision, speech, and multimodal capabilities of GPT-4o. The study employs standardized exam questions, reasoning tasks, and translation assessments to assess the model's language capability. Additionally, GPT-4o's vision and speech capabilities are tested through image classification and object recognition tasks, as well as accent classification. The multimodal evaluation assesses the model's performance in integrating visual and linguistic data. Our findings reveal that GPT-4o demonstrates high accuracy and efficiency across multiple domains in language and reasoning capabilities, excelling in tasks that require few-shot learning. GPT-4o also provides notable improvements in multimodal tasks compared to its predecessors. However, the model shows variability and faces limitations in handling complex and ambiguous inputs, particularly in audio and vision capabilities. This paper highlights the need for more comprehensive benchmarks and robust evaluation frameworks, encompassing qualitative assessments involving human judgment as well as error analysis. Future work should focus on expanding datasets, investigating prompt-based assessment, and enhancing few-shot learning techniques to test the model's practical applicability and performance in real-world scenarios.

With advancements in digital technologies such as 5G communications, big data, and cloud computing, the components of network operation systems have become increasingly complex, significantly complicating system monitoring and maintenance. Correspondingly, automated log anomaly detection has become a crucial means to ensure stable network operation and protect them from malicious attacks or failures. Conventional machine learning and deep learning methods assume consistent distributions between the training and testing data, adhering to a closed-set recognition paradigm. Nevertheless, in realistic scenarios, systems may encounter new anomalies that were not present in the training data, especially in log anomaly detection. Inspired by evidential learning, we propose a novel anomaly detector called LogEDL, which supervises training of the model through an evidential loss function. Unlike traditional loss functions, the evidential loss function not only focuses on correct classification, but also quantifies the uncertainty of predictions. This enhances the robustness and accuracy of the model in handling anomaly detection tasks, while achieving functionality similar to open-set recognition. To evaluate the proposed LogEDL method, we conduct extensive experiments on three datasets, i.e., HDFS, BGL, and Thunderbird, to detect anomalous log sequences. Experimental results demonstrate that our proposed LogEDL achieves state-of-the-art performance in anomaly detection.

COVID-19 was a terrible worldwide pandemic that caused a global public health crisis, with numerous deaths and a severe economic depression. To suppress the spread of COVID-19 and initiate patient isolation and contact tracing to reduce its effect, early diagnosis was required, with real-time reverse transcriptase polymerase chain reaction (RT-PCR) swab test being the most widely used. However, the RT-PCR test was sometimes found to be lengthy and inaccurate and; for that reason, in many cases of severe complications of COVID-19 pneumonia, chest screening with radiography imaging was preferred. In this paper, we carried out a study for COVID-19 detection from computed tomography images, and compared the obtained results using deep learning (DL) and support vector machine (SVM). The findings showed that, despite the excellent results shown by deep learning, the accuracy in predicting COVID-19 using SVM, in the presence of small databases, was slightly higher than DL. The execution time using SVM was also shorter

This research introduces a hybrid feature extraction approach that combines Linear Discriminant Analysis (LDA) and Multilayer Perceptron (MLP) methods to address the challenges of reducing feature vector dimensionality and accurately classifying smartphone-based human activities. Moreover, to refine activity classification accuracy, Support Vector Machine (SVM) optimization with Stochastic Gradient Descent (SGD) is employed. LDA, a statistical tool, is leveraged to derive a new feature space for data projection, enhancing class separation and test feature label prediction. The proposed approach, named LMSS, was evaluated using the UCI-HAR dataset and compared with state-of-the-art models. The results demonstrate that the proposed approach outperformed the best-performing method over this dataset. It achieved an accuracy rate of 99.52%, precision of 99.55%, recall of 99.53%, and an F1-score of 99.54%, highlighting the effectiveness of the proposed method in accurately classifying the data.

Our world is increasingly challenged by managing the impacts of natural disasters, particularly floods, which are frequent, dangerous, and costly. Traditional flood mapping methods, reliant on Optical satellites (MSI), struggle under cloud cover which is typical during such events. Synthetic Aperture Radar (SAR) offers a promising alternative with its cloud-penetrating capability, though its use has been limited due to complexity and data labeling challenges. This project aims to develop a SAR-based flood segmentation model that can rapidly and accurately map floods globally, with high adaptability to different flood types and regions. By utilizing deep learning and a novel transfer-learning technique that combines the strengths of Optical/MSI and SAR data, the model seeks to bypass the challenges of manual labeling and improve mapping accuracy. Initial results show the model's effective generalization across various flood events, with superior performance indicated by an Intersection over Union (IoU) of 0.72, outperforming existing methods and demonstrating promising capabilities in precise flood mapping.

Accurate crop diseases classification is crucial for ensuring food security and enhancing agricultural productivity. However, existing crop disease classification algorithms primarily focus on a single image modality and typically require a large number of samples. Our research counters these issues by using pre-trained Vision-Language Models (VLMs), which enhances multimodal synergy for better crop disease classification than traditional unimodal approaches. Firstly, we apply the multimodal model Qwen-VL to generate meticulous textual descriptions for representative disease images selected through clustering from the training set, which will serve as prompt text for generating classifier weights. Compared to using solely the language model for prompt text generation, this approach better captures and conveys fine-grained and image-specific information, thereby enhancing prompt quality. Secondly, we integrate the cross-attention and the SE (Squeeze-and-Excitation) attention into the training-free mode VLCD (Vision-Language model for Crop Diseases classification) and the training-required mode VLCD-T (VLCD-Training) respectively for prompt text processing, enhancing classifier weights by emphasizing key text features. Experimental outcomes conclusively prove our method’s heightened classification effectiveness in few-shot crop disease scenarios, tackling data limitations and intricate disease recognition issues. It offers a pragmatic tool for agricultural pathology and reinforces the smart farming surveillance infrastructure.

This paper explores the utilization of ensemble models and learning analytics techniques to predict student academic performance. With the advent of educational big data, institutions are increasingly leveraging advanced analytics to gain insights into student learning patterns and optimize educational outcomes. Ensemble models, which combine the predictive power of multiple algorithms, offer a robust approach to enhance prediction accuracy. The performance of the ensemble models was analyzed and compared using the Open University Learning Analytics Dataset, which consists of sources such as demographic information, historical performance data, and engagement metrics for 23,344 students. The evaluation of various ensemble models across different classification scenarios revealed that the proposed stacking model consistently emerges as the best-performing model, excelling in both multi-class and binary classification tasks.

Presence detection is essential in many applications, including smart homes, office automation, and public space management. Traditional technologies like infrared sensors and cameras frequently encounter obstacles due to privacy concerns, high costs, and complexity. The study provides a novel way to detect presence based on Wi-Fi signals and machine learning algorithms, utilizing the ESP8266 microcontroller. The suggested system attempts to reliably detect the presence of people in a given place by evaluating fluctuations in Wi-Fi signal strength and patterns. The theoretical basis of this strategy is investigated, including the ESP8266’s capabilities for Wi-Fi data gathering and the use of machine learning methods for data analysis. The methodology includes data collecting with the ESP8266, preprocessing, feature extraction, and training and evaluating machine learning models. This system’s potential uses include smart home automation, office occupancy tracking, and crowd management in public settings. The paper also discusses potential issues such as signal interference, data privacy concerns, and the limitations of Wi-Fi detection. Future work is also mentioned, including the suggested system’s installation and testing and enhancement proposals to improve accuracy and scalability. This research aims to provide a cost-effective, non-intrusive, and scalable presence detection system by harnessing the power of Wi-Fi signals and machine learning, enhancing intelligent environments

Underwater imaging presents unique challenges, notably color distortions and reduced contrast due to light attenuation and scattering. Most underwater image enhancement methods, first use linear transformations for color compensation and then enhance the image. We observed that linear transformation for color compensation is not suitable for certain images. For such images, a non-linear mapping is a better choice. Our paper introduces a unique underwater image restoration approach leveraging a streamlined convolutional neural network (CNN) for dynamic weight learning for linear and non-linear mapping. In the first phase, a classifier is applied that classifies the input images into Type-I or Type-II. In the second phase, depending on the classification, the Deep Line Model (DLM) for Type-I images or the Deep Curve Model (DCM) for Type-II images. For mapping an input image to an output image, DML creatively combines color compensation and contrast adjustment in a single step and uses deep lines for transformation, whereas DCM employs higher-order curves. Both models utilize lightweight neural networks that learn per-pixel dynamic weights based on the input image’s characteristics. Comprehensive evaluations on benchmark datasets using metrics like peak signal-to-noise ratio(PSNR) and root mean squares error (RMSE) affirm our method’s effectiveness in accurately restoring underwater images, outperforming existing techniques.

This study is based on the BP neural network regression model optimised by genetic algorithm and dedicated to predicting the amount of car purchase, this study is important for all parties in the automotive industry. By plotting the violins of six indicators, including gender, age, annual salary, credit card debt, net worth, and car purchase amount, we found that, except for gender, the other data showed a trend of dense in the middle gradually decreasing to the upper and lower sides, with an overall normal distribution. The correlation analysis shows that the amount of car purchase has a high positive correlation with age (0.63), followed by annual income (0.62), the correlation between net worth and the amount of car purchase is 0.49, while the correlation between credit card debt and the amount of car purchase is low (0.03). In the experimental part, we analyse the predicted-actual values of the training and test sets in comparison and find that the model predicted values are very close to the actual values. The RMSE of the training set is 11.559 and the RMSE of the test set is 11.4835, which is not much different from each other, indicating that the model has good generalisation ability. In addition, the R2 of both the training and test sets is 1, indicating that the model fits well. The MAE of the training set is 8.2512 and the MAE of the test set is 9.578; the MBE of the training set is -4.3035 and the MBE of the test set is -2.4547.These metrics indicate that the model performs well in predicting the amount of car purchase. In conclusion, this study provides an effective and reliable method to predict the amount of car purchase, creating a useful bridge between car manufacturers, dealers and consumers.

Every year, the Montreal police are confronted with countless crimes committed by criminals. These crimes affect the quality of life of city residents and impose a socio-economic burden on the city. In this study, we conduct a comparative analysis based on several machine learning algorithms to develop a model to predict the crime category in Montreal. The performance of algorithms such as eXtreme Gradient Boosting (XGBoost), Decision Trees (DT) and Random Forest (RF) were analyzed. The performance analysis takes into account the performance metrics such as precision, accuracy, recall and F1-score. This analysis was based on crime data in Montreal from 2015 to 2023. This data is characterized by a strong imbalance between crime categories. To address the data imbalance problem, a data balancing approach based on the SMOTE-ENN algorithm was adopted. In the exploratory data analysis phase, temporal trends by crime category were highlighted. The results of the analysis showed that the XGBoost algorithm outperformed the other two. Specifically, the XGBoost algorithm achieves an accuracy of 92%, while DT and RF achieve an accuracy of 86% and 84%, respectively. As a result, XGBoost was deployed via a web application using the Flask and Swagger UI Python frameworks. This study provides the Montreal police with an effective tool to better utilize their resources in fighting crime. In addition, policymakers in the city of Montreal can use this tool to identify high-risk areas and give them more attention.

Given the widespread use of AI language models such as ChatGPT, there is an urgent need 1 to understand their political biases in different languages. This study aims to uncover such biases. To 2 investigate this issue, we applied IDRLabs’ Political Coordinates Test to ChatGPT based on GPT-4 3 in four languages - English, French, German, and Italian. The results confirm previous studies that 4 identified a left-libertarian bias in ChatGPT. There was remarkable consistency in political orientation 5 across all four languages, although minor variations were found. Differences in ChatGPT response 6 distributions across languages could indicate culturally determined differences in interpretation 7 of political questions. The results highlight the need to diversify and optimize data collection and 8 model training methods to ensure greater fairness and neutrality and minimize the influence of bias. 9 Our study highlights the importance of cultural differences in the processing and interpretation of 10 political information by AI models.

The challenging terrain and deep ravines characteristic of mountainous regions often result in slower path planning and suboptimal flight paths for Unmanned Aerial Vehicles (UAVs) when traditional meta-heuristic optimization algorithms are employed. This study investigates the application of an Improved Enhanced Snake Optimizer (IESO) for three-dimensional path planning in a simulated rugged mountainous terrain. The initialization process in the enhanced snake optimizer is refined by integrating the Chebyshev chaotic map. Additionally, a non-monotonic factor is introduced to modulate the "temperature," and a boundary condition is incorporated into the dynamic opposition learning mechanism. These modifications collectively reduce the likelihood of population convergence to local optima during optimization. The feasibility of IESO is validated through time complexity and global convergence analyses. Comparative simulation experiments benchmarked IESO against five state-of-the-art optimization algorithms across test functions and path-planning simulated scenarios. The experimental results indicate that the enhanced algorithm achieves superior optimization precision, faster convergence speeds, and improved quality in the planned trajectories by 30%.

This paper presents an advanced approach for the Detection of MRI brain tumor images using Convolutional Neural Networks (CNNs). The study focuses on employing sophisticated techniques such as data augmentation, model optimization, and rigorous performance evaluation metrics to achieve high accuracy in multi-class classification tasks. Additionally, we integrate the concept of Self-Expanding Convolutional Neural Networks (SECNNs), which dynamically adjust model complexity during training. This approach ensures optimal performance while maintaining computational efficiency. The dataset used in this research comprises MRI images categorized into four distinct classes: glioma, meningioma, no tumor, and pituitary tumor. Our method demonstrates a notable test accuracy of 93.98%

Dynamic sparse training is an effective strategy to alleviate the training and inference demands of artificial neural networks. However, current sparse training methods face a challenge in achieving high levels of sparsity while maintaining performance comparable to that of their fully connected counterparts. The Cannistraci-Hebb training (CHT) method produces an ultra-sparse advantage compared to fully connected training in various tasks by using a gradient-free link regrowth method, which relies solely on the network topology. However, its rigid selection based on link prediction scores may lead to epitopological local minima, especially at the beginning of the training process when the network topology might be noisy and unreliable. In this article, we introduce the Cannistraci-Hebb training soft rule (CHTs), which applies a flexible approach to both the removal and regrowth of links during training, fostering a balance between exploring and exploiting network topology. Additionally, we investigate the network topology initialization using several approaches, including the bipartite scale-free and bipartite small-world network models. Empirical results show that CHTs can surpass the performance of fully connected networks with MLP architecture by using only 1% of the connections (99% sparsity) on the MNIST, EMNIST, and Fashion MNIST datasets and can provide remarkable results with only 0.1% of the links (99.9% sparsity). In some MLPs for image classification tasks, CHTs can reduce the active neuron network size to 20% of the original nodes (neurons), demonstrating a remarkable ability to generalize better than fully connected architectures, reducing the entire model size. This represents a relevant result for dynamic sparse training. Finally, we present evidence from larger network models such as Transformers, with 10% of the connections (90% sparsity), where CHTs outperform other prevalent dynamic sparse training methods in machine translation tasks.

Recently, a significant number of tools for detecting dangerous objects have been developed. Unfortunately, the performance offered by them is overestimated due to the poor quality of the datasets used (insufficiently numerous, contain items not strictly related to dangerous objects, insufficient range of presentation conditions). To fill in a gap in this area we have built an extensive dataset dedicated to detecting the objects most often used in various acts of breaching public security (baseball bat, gun, knife, machete, rifle). This collection contains images presenting the detected objects with different quality and under different environmental conditions. We believe that the results obtained from it are more reliable and give a better idea of the detection accuracy that can be achieved under real conditions. We used the Faster R-CNN with different backbone networks in the study. The best results were obtained for the ResNet152 backbone. The mAP value was 85%, while the AP level ranged from 80% to 91%, depending on the item detected. An average real-time detection speed was 11-13 FPS. Both the accuracy and speed of the Faster R-CNN model allow it to be recommended for use in public security monitoring systems aimed at detecting potentially dangerous objects.

The enhanced multi-objective symbolic discretization for time series (eMODiTS) method employs a flexible discretization scheme using different value cuts for each non-equal time interval, which requires a computational cost for evaluating each objective function. Therefore, surrogate models were implemented to minimize this disadvantage. Nevertheless, each solution found by eMODiTS is a different-sized vector, so the surrogate model must be able to handle data sets under this characteristic. Consequently, this work's contribution lies in analyzing the surrogate models' implementation on the time series discretization, where each potential scheme is a real-number different-sized vector. For this reason, the surrogate model proposed was k-nearest Neighbors for regression with Dynamic Time Warping as a distance measure. Results suggest our proposal finds a suitable approximation to the final eMODiTS solutions with a functions evaluation reduction rate between 15% and 95%. However, according to Pareto front performance measures, the proposal Pareto front is competitive compared to the eMODiTS Pareto front, reaching an average Generational Distance (GD) between 0.0447 and 0.0536. Moreover, the average Hypervolume Ratio (HVR) ranges between 0.334 and 0.3891. Finally, our proposal compared against SAX-based methods presents a similar behavior regarding classification tasks and statistical tests.

Nonlinear optimization (NOPT) is a meaningful tool for solving complex tasks in fields like engineering, economics, and operations research, among others. However, NOPT has problems when it comes to dealing with data variability and noisy input measurements that lead to incorrect solutions. Furthermore, nonlinear constraints may result in outcomes that are either infeasible or suboptimal, such as non-convex optimization. This paper introduces a novel regularized physics-informed neural network (RPINN) framework as a new NOPT tool for both supervised and unsupervised data-driven scenarios. By using custom activation functions and regularization penalties in an artificial neural network (ANN), RPINN can handle data variability and noisy inputs. Besides, it uses physics principles to build the network architecture, computing the optimization variables based on network weights and learned features. In addition, it uses automatic differentiation training to make the system scalable and cut down on computation time through batch-based back-propagation. The test results for both supervised and unsupervised NOPT tasks show that our RPINN can provide solutions that are competitive compared to state-of-the-art solvers. In turn, RPINN’s robustness against noisy input measurements makes it particularly valuable in environments with fluctuating information. Additionally, RPINN’s ANN-based foundation offers significant flexibility and scalability.

Generating executable code from natural language instructions using Large Language Models (LLMs) presents challenges such as semantic understanding and handling ambiguous input. This study focuses on the SantaCoder model and explores the impact of in-context learning on code generation using the MBPP and HumanEval datasets for evaluation. Our results demonstrate significant improvements in three key metrics (defined in the paper): correctness@k, similarity@k and pass@k. To address the problem of selecting optimal demonstrations to maximize correctness and pass rates, we investigate two methods: latent concept selection and random selection in this paper. These findings highlight the effectiveness of in-context learning and the critical role of demonstration selection in enhancing the accuracy, efficiency, and versatility of the SantaCoder model in code generation.

Mulberry leaves, when infected by pathogens, can suffer significant yield loss or even death if early disease detection and timely spraying are not performed. To enhance the detection performance of mulberry leaf diseases in natural environments and to precisely locate early small lesions, we propose a high-precision, high-efficiency disease detection algorithm named YOLOv8-RFMD. Based on improvements to You Only Look Once version 8 (YOLOv8), we first proposed the Multi Dimension Feature Attention (MDFA) module, which integrates important features at the pixel-level, spatial and channel dimensions. Building on this, we designed the RFMD Module, which consists of the Conv-BatchNomalization-SiLU (CBS) module, Receptive-Field Coordinated Attention (RFCA) Conv, and MDFA, replacing the Bottleneck in the model’s Residual block. We then employed Adown down sampling structure to reduce the model size and computational complexity. Finally, to improve the detection precision of small lesion features, we replaced the complete intersection over union (CIOU) loss function with the Normalized Wasserstein Distance (NWD) loss function. Results show that the YOLOv8-RFMD model achieves an mAP50 of 94.3% and an mAP50:95 of 67.8% on experimental data, representing increases of 2.9% and 4.3% respectively compared to the original model. The model size is reduced by 0.53 MB to just 5.45 MB, and the number of floating-point operations is reduced by 0.3 G to only 7.8 G. The improved model meets the deployment requirements of mobile embedded devices and can provide a theoretical reference for the automated spraying operations of mulberry leaves.

：Non-Euclidean data, such as social networks, and citation relationships between documents, has node information and structural information. Graph Convolutional Network(GCN) can automatically learn node features and association information between nodes. The core ideology of the graph convolutional network is to aggregate node information by using edge information, thereby generating a new node feature. In the process of updating node features, there are two core influencing factors. One is the number of neighboring nodes of the central node, the other is the contribution of the neighboring nodes to the central node. Due to the previous GCN methods not simultaneously considering the numbers and different contributions of neighboring nodes to the central node, we design the Adaptive Attention Mechanism(AAM). To further enhance the representational capability of the model, we utilize Multi-Head Graph Convolution(MHGC). Finally, we adopt cross-entropy(CE) loss function to describe the difference between the predicted results of node categories and the ground truth (GT). Combined with backpropagation, this ultimately achieves accurate node classification. Based on AAM, MHGC and CE, we contrive the novel Graph Adaptive Attention Network (GAAN). Experiments show that the classification accuracy has achieved outstanding performances on Cora, Citeseer and Pubmed datasets.

Evolutionary algorithms and swarm intelligence algorithms find applicability in reinforcement learning of neural networks due to their independence from gradient-based methods. To achieve successful training of neural networks using these algorithms, careful considerations must be made to select appropriate algorithms due to the availability of various algorithmic variations. In Part1, the author previously reported experimental evaluations on Evolution Strategy　for reinforcement learning of binary neural networks, utilizing the Pendulum control task. This article constitutes Part2 of the series of comparative research. In this study, Genetic Algorithm is adopted as another evolutionary algorithm. Wilcoxon signed rank test revealed that there was no statistically significant difference between the fitness scores obtained using GA and those using ES. However, the p-value 0.11 indicated that GA worked better than ES on this training task. As the values of binary weights, {-1, 1} were significantly superior to {0, 1} (p < .01). The motion of the pendulum controlled by the binary MLP after the training showed that the binary MLP successfully swung the pendulum swiftly into an inverted position and maintained its stability after inversion.

The increasing concern regarding climate change continues to motivate research in the Arctic. Within the delicate cycle of the cryosphere, leads are an important kinematic feature that regulate heat balances in the Arctic. Therefore, it is necessary to quantify the genesis of leads over time to identify when and where changes are occurring. The use of a principal component analysis (PCA) is one such tool that is used to identify the characteristics of seasonal ice variability. The PCA not only identifies “normal” or “dominant” conditions or features, but also extract anomalous spatial features from an archive of long-term sequence of images.

The capacitated multi-item lot-sizing problem abbreviated as CLSP, is to determine the lot sizes of products in each period in a given planning horizon of finite periods meeting the product demand and resource limits in each period, and minimize the total cost consisting of production, inventory holding, and setup costs. CLSPs are often encountered in industry production settings and thus the solution is significant. In this paper, we propose a Lagrange relaxation (LR) approach for the solution of its. The approach decomposes the CLSP into several uncapacitated single-item lot-sizing problems each of which can be easily solved by dynamic programming. The feasible solutions are achieved by solving the resulting transportation problems with a proposed stepping-stone algorithm and a fix-up procedure. The Lagrange multipliers are updated by using subgradient optimization. Experimental results show that the LR approach explores high-quality solutions and has better applicability compared with other commonly used solution approaches in the literature.

This study presents PlanteSaine, a novel mobile application powered by Artificial Intelligence (AI) models explicitly designed for maize, tomato, and onion farmers in Burkina Faso. Agriculture in Burkina Faso, like many developing nations, faces substantial challenges from plant pests and diseases, posing threats to both food security and economic stability. PlanteSaine addresses these challenges by offering a comprehensive solution that provides farmers with real-time identification of pests and diseases. Farmers capture images of affected plants with their smartphones, and PlanteSaine's AI system analyzes these images to provide accurate diagnoses. The application's offline functionality ensures accessibility even in remote areas with limited internet connectivity, while its messaging feature facilitates communication with agricultural authorities for guidance and support. Additionally, PlanteSaine includes an emergency alert mechanism to notify farmers about pest and disease outbreaks, enhancing their preparedness to deal with these threats. An AI-driven framework, featuring an image feature extraction phase with EfficientNetB3 and an artificial neural network (ANN) classifier, was developed and integrated into PlanteSaine. The evaluation of PlanteSaine demonstrates its superior performance compared to baseline models, showcasing its effectiveness in accurately detecting diseases and pests across maize, tomato, and onion crops. Overall, this study highlights the potential of PlanteSaine to revolutionize agricultural technology in Burkina Faso and beyond. Leveraging AI and mobile computing, PlanteSaine provides farmers with accessible and reliable pest and disease management tools, ultimately contributing to sustainable farming practices and enhancing food security. The success of PlanteSaine underscores the importance of interdisciplinary approaches in addressing pressing challenges in global agriculture

This paper focuses on enhancing the object detection and decision-making capabilities of Autonomous Emergency Medical Response Systems (AEMRS). Using advanced sensors, Convolutional Neural Networks (CNN), and reinforcement learning, we propose a model that processes environmental data to identify obstacles and make optimal navigation decisions. The integration of these technologies aims to minimize response time and improve the efficiency of emergency medical services. The perception system utilizes radar, LiDAR, and camera inputs to create a comprehensive understanding of the environment. These data are processed through a series of CNN layers to detect and classify objects such as vehicles, pedestrians, and road signs. On the decision-making front, a Q-learning algorithm is employed to enable the ambulance to learn from its interactions with the environment, continuously improving its route planning and collision avoidance strategies. By combining these advanced AI techniques, the proposed AEMRS can significantly enhance the speed and reliability of emergency responses, ultimately saving more lives. This paper presents the design, implementation, and simulation results of the proposed system, demonstrating its potential to revolutionize emergency medical services.

This paper presents our approaches for the SMM4H’24 Shared Task 5 on the binary classification of English tweets reporting children’s medical disorders. Our first approach involves fine-tuning a single RoBERTa-large model, while the second approach entails ensembling the results of three fine-tuned BERTweet-large models. We demonstrate that although both approaches exhibit identical performance on validation data, the BERTweet-large ensemble excels on test data. Our best-performing system achieves an F1-score of 0.938 on test data, outperforming the benchmark classifier by 1.18%.

Traffic accidents are among the most common causes of death worldwide. According to statistics from the World Health Organization (WHO), 50 million people are involved in traffic accidents every year. Canada, particularly Montreal, is not immune to this problem. Data from the Société de l’Assurance Automobile du Québec (SAAQ) shows that there were 392 deaths on Québec roads in 2022, 38 of them related to the city of Montreal. This value represents an increase of 29.3% for the city of Montreal compared to the average for the years 2017 to 2021. In this context, it is important to take concrete measures to improve traffic safety in the city of Montreal. In this article, we present a web-based solution based on machine learning that predicts the severity of traffic accidents in the city of Montreal. This solution uses dataset of traffic accidents that occurred in Montreal between 2012 and 2021. Classification algorithms such as eXtreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), Random Forest (RF) and Gradient Boosting (GB) were used to develop the prediction model. When evaluating the prediction model, performance metrics such as precision, recall, F1 score, and accuracy are taken into account. The performance analysis shows an excellent accuracy of 96% for the prediction model based on the XGBoost classifier. The other models (CatBoost, RF, GB) achieved 95%, 93% and 89% accuracy, respectively. The prediction model based on the XGBoost classifier was deployed using a client-server web application managed by Swagger-UI and the Flask Python framework.