Artificial Intelligence (AI) is becoming increasingly important, especially in the medical field. While AI has been used in medicine for some time, its growth in the last decade has been remarkable. Specifically, Machine Learning (ML) and Deep Learning (DL) techniques in medicine have been increasingly adopted thanks to the growing abundance of health-related data, improved suitability of such techniques for managing large data-sets, and more computational power. The Italian scientific community has been instrumental in advancing this research area. This article aims to conduct a comprehensive investigation of the ML and DL methodologies and applications used in medicine by the Italian research community in the last five years.

Over the past decade, Deep Learning (DL) had been applied in a large number of optical sensors’ applications. DL algorithms can improve the accuracy and reduce the noise level in optical sensors. Optical sensors are considered as promise technology for the modern intelligent sensing platforms. These sensors are widely used to process monitoring, quality prediction, pollution, defense, security and many other applications. However, they suffer major challenges such as the large generated data and low processing speed for that data and moreover the much cost of these sensor. These challenges can be mitigated by integrating deep learning system with the optical sensor technologies. This paper presents recent studies that integrate DL algorithms with optical sensors applications. This paper also highlights several of DL algorithms directions that promise a considerable impact on use for optical sensor applications. Moreover, this study provides new directions for the future related research development.

Hundreds of millions of people on the planet are affected by malnourishment. This contributes to the vulnerability of large swaths of the population, especially children under five years old extremely vulnerable to diseases and even death in the least developed countries. Today, by providing a substantial share of global protein intake, as well as fatty acids and micronutrients, fisheries contribute to global food security. As fish stocks in the upper sea levels have already been over-exploited, there might not be enough in supplying food security, deep-sea fisheries have been increasingly discussed to address this problem. Some mesopelagic fishes show the potential to provide important nutrients. Another way of supplying food security might be through using mesopelagic fish as fish feed. However, fishing in the mesopelagic zone could lead to severe ecological problems and the impact on the biological carbon pump is uncertain. This paper highlights and juxtaposes different perspectives regarding exploitation pathways of the fish riches of deep seas and reviews best practice model projects that deal with uncertainties related to fishery management in the mesopelagic zone.

This study compares the performance of existing studies on multimodal emotion recognition, and proposes a model that fuses two modalities with the speaker's text and voice signals as input values and detects depression. Based on the DAIC-WOZ dataset, voice features were extracted using CNN, text features were extracted using Transformers, and two modalities were fused through a tensor fusion network. We also build a model to detect whether the speaker is depressed or not using LSTM in the final layer. This study suggests the possibility of increasing access to mental illness diagnosis by enabling patients to detect depression on their own in daily conversations. If the model proposed in this study is developed and the voice conversation system is connected, it will be easier for patients who cannot visit the hospital periodically or who are reluctant to visit the hospital to check their condition and seek recovery. Furthermore, it can be expanded to multi-label classification for various mental diseases and used as a simple self-mental disease diagnosis tool.

Abstract: Image aesthetic assessment (IAA) with neural attention has made significant progress due to its effectiveness in object recognition. Current studies have shown that the features learned by convolutional neural networks (CNN) at different learning stages indicate meaningful information. The shallow feature contains the low-level information of images and the deep feature perceives the image semantics and themes. Inspired by this, we propose a visual enhancement network with feature fusion (FF-VEN). It consists of two sub-modules, the visual enhancement module (VE module) and the shallow and deep feature fusion module (SDFF module). The former uses an adaptive filter in the spatial domain to simulate human eyes according to the region of interest (ROI) extracted by neural feedback. The latter not only takes out the shallow feature and the deep feature by transverse connection, but also uses a feature fusion unit (FFU) to fuse the pooled features together with the aim of information contribution maximization. Experiments on standard AVA dataset and Photo.net dataset show the effectiveness of FF-VEN.

Urban scene point cloud pose significant challenges for registration due to its large data volume, similar scenarios and dynamic objects. In this paper, we propose PCRMLP, a model for urban scene point cloud registration that achieves comparable registration performance to prior learning-based methods. Compared to previous works which focus on extracting features and estimating correspondence, the model estimates the transformation implicitly from concrete instances. An instance-level urban scene representation method is introduced to extract instance descriptors via semantic segmentation and DBSCAN, which enable the model to obtain robust instance features, filter dynamic objects and estimate transformation in a more logical manner. Then a lightweight network consisting of MLPs is employed to obtain transformation in an encoder-decoder manner. We validate the approach on KITTI dataset. Experimental results demonstrate that PCRMLP can obtain a satisfactory coarse transformation from instance descriptors just in 0.0028s. With a subsequent ICP refinement module, the proposed method achieves higher registration accuracy and computational efficiency than prior learning-based works.

Estimating plume cloud height is essential for various applications, such as global climate models. Smokestack plume rise is the constant height at which the plume cloud is carried downwind as its momentum dissipates and the plume cloud and the ambient temperatures equalize. Although different parameterizations are used in most air-quality models to predict the plume rise, they have been unable to estimate it properly. This paper proposes a novel framework to monitor smokestack plume clouds and make long-term, real-time measurements of the plume rise. For this purpose, a three-stage framework is developed based on Deep Convolutional Neural Networks (DCNNs). In the first stage, an improved Mask R-CNN, called Deep Plume Rise Network (DPRNet), is applied to recognize the plume cloud. Then, image processing analysis and least squares theory are respectively used to detect the plume cloud’s boundaries and fit an asymptotic model into their centerlines. The y-component coordinate of this model’s critical point is considered the plume rise. In the last stage, a geometric transformation phase converts image measurements into real-life ones. A wide range of images with different atmospheric conditions, including day, night, and cloudy/foggy, have been selected for the DPRNet training algorithm. Obtained results show that the proposed method outperforms widely-used networks in smoke border detection and recognition.

The increasing use of deep learning techniques to manipulate images and videos, commonly referred to as "deepfakes," is making more and more challenging to differentiate between real and fake content. While various deepfake detection systems have been developed, they often struggle to detect deepfakes in real-world situations. In particular, these methods are often unable to effectively distinguish images or videos when these are modified using novel techniques which have not been used in the training set. In this study, we carry out an analysis of different deep learning architectures in an attempt to understand which is more capable of better generalizing the concept of deepfake. According to our results, it appears that Convolutional Neural Networks (CNNs) seem to be more capable of storing specific anomalies and thus excel in cases of datasets with a limited number of elements and manipulation methodologies. The Vision Transformer, conversely, is more effective when trained with more varied datasets, achieving more outstanding generalization capabilities than the other methods analysed. Finally, the Swin Transformer appears to be a good alternative for using an attention-based method in a more limited data regime. All the analyzed architectures seem to have a different way to look at deepfakes but since in a real-world environment, the generalization capability is essential, based on the carried out experiments the Vision Transformer seems to provide superior performances.

Deep learning has already revolutionised the way we process a wide range of data, in many areas of our daily life. The ability to learn abstractions and relationships from heterogeneous data, has provided impressively accurate prediction and classification tools to handle increasingly big datasets. This has a significant impact on the growing wealth of omics datasets, with the unprecedented opportunity for a better understanding of the complexity of living organisms. While this revolution is transforming the way we analyse these data, explainable deep learning is emerging as an additional tool with the potential to change the way we interpret biological data. Explainability addresses critical issues such as transparency, so important when computational tools are introduced especially in clinical environments. Moreover, it empowers artificial intelligence with the capability to provide new insights in the input data, thus adding an element of discovery to these already powerful resources. In this review we provide an overview of the transformative effects explainable deep learning is having on multiple sectors, ranging from genome engineering and genomics, from radiomics to drug design and clinical trials. We offer a perspective to life scientists, to better understand the potential of these tools, and a motivation to implement them in their research, by suggesting learning resources they can use to move their first steps in this field.

Acanthopanax senticosus(rupr. rt. Maxim.; AS) is a medicinal plant used in the clinical treatment of cerebrovascular diseases and central nervous system disorders, significantly improving blood lipid levels and endothelial cell function in patients with acute cerebral infarction. Isofraxidin, one of the active ingredients of AS, is the core of the medical effects, and its extraction and purification depend on organic solvents. Deep eutectic solvents (DESs) is a new green solvent, synthesized by intermolecular hydrogen bonding between hydrogen bond donors (HBD) and hydrogen bond acceptors (HBA), which has non-toxic, high separation and purification efficiency, and environmentally friendly compared with traditional organic solvents. In this paper, DES was used for the extraction of isofraxidin from AS. The primary findings demonstrate that the DESs had a viscosity higher than that of ethanol, and even adding a tiny amount of water (about 10%) would trigger solvent redistribution, leading to a considerable reduction in solvent viscosity. In comparison to ethanol, the extraction rate of isofraxidin by DESs was 2-3 times higher. Thus, this work developed a new technique for using green extraction of isofraxidin that has some practical implications.

Initial choice of Learning Rate is a key part of gradient based methods and has a great effect on the performance of the Deep Learning Model.This paper studies the behavior of multiple gradient based optimization algorithm which are commonly used in Deep Learning and compare their performance on various learning rate. As observed popular choice of optimization algorithms are highly sensitive to various choice of learning rates. Our goal is to find which optimizer has an edge over others for a specific setting. We look at two datasets namely MNIST and CIFAR10 for benchmarking. The results are quite surprising, and it will help us to choose a learning rate more efficiently.

Face detection is an important problem in computer vision because it enables a wide range of applications, such as facial recognition and analysis of human behavior. The problem is challenging because of the large variations in facial appearance across different individuals and different lighting and pose conditions. One way to detect faces is to utilize a highly advanced face detection method, such as RetinaFace, which uses deep learning techniques to achieve high accuracy in various datasets. However, even the best face detectors can produce false positives, which can lead to incorrect or unreliable results. In this paper, we propose a method for reducing false positives in face detection by using information from a depth map. A depth map is a two-dimensional representation of the distance of objects in an image from the camera. By using the depth information, the proposed method is able to better differentiate between true faces and false positives. The authors evaluate their method on a combined dataset of 549 images, containing a total of 614 upright frontal faces. The results show that the proposed method is able to significantly reduce the number of false positives without sacrificing the overall detection rate. This indicates that the use of depth information can be a useful tool for improving face detection performance.

Tardive dyskinesia (TD) is a phenomenon predominantly observed as a result of the long-term use of dopamine receptor blockers (antipsychotics). TD is a group of involuntary, irregular hyperkinetic movements, mainly in the muscles of the face, eyelid muscles, lips, tongue, and cheeks, and less frequently in the limbs, neck, pelvis, and trunk. In some patients, TD takes on an extremely severe form, massively disrupting functioning, moreover, causing stigmatization and suffering. Deep brain stimulation (DBS), a method used, among others, in Parkinson's disease, is also an effective treatment for TD and often becomes a method of last resort, especially in severe, drug-resistant forms. The procedure is relatively new in TD, so the available reliable clinical studies are few and consist mainly of case reports. Unilateral and bilateral stimulation of two sites has proven efficacy in TD treatment. Most authors describe stimulation of the globus pallidus internus (GPi); less frequent descriptions involve the hypothalamic nucleus (STN). In the present paper, we provide up-to-date information on the stimulation of both mentioned brain areas. We also compare the efficacy of the two methods by comparing the two available studies, which included the largest groups of patients. Although GPi stimulation is more frequently used clinically, our initial analysis indicates comparable results (reduction of involuntary movements) with STN DBS.

Deep Learning is an extremely important research topic in Earth Observation. Current use-cases range from semantic image segmentation, object detection to more common problems found in computer vision such as object identification. Earth Observation is an excellent source for different types of problems and data for Machine Learning in general and Deep Learning in particular. It can be argued that both Earth Observation and Deep Learning as fields of research will benefit greatly from this recent trend of research. In this paper we take several state of the art Deep Learning network topologies and provide a detailed analysis of their performance for semantic image segmentation for building footprint detection. The dataset used is comprised of high resolution images depicting urban scenes. We focused on single model performance on simple RGB images. In most situations several methods have been applied to increase the accuracy of prediction when using deep learning such as ensembling, alternating between optimisers during training and using pretrained weights to bootstrap new models. These methods although effective, are not indicative of single model performance. Instead, in this paper, we present different topology variations of these state of the art topologies and study how these variations effect both training convergence and out of sample, single model, performance.

If the development of machine learning and artificial intelligence plays a role in many fields of research and technology today, it has a special relationship with neurosciences. Indeed, historically inspired by our knowledge of the brain, deep learning shares some vocabularies with neurosciences and can sometimes be considered a brain’s model. Taking the particular example of epileptic seizure, which can develop in any biological neural tissue, we raise the question if and how the models used for deep learning can capture or model these pathological events. This particular example is a starting point to discuss the nature, limits, and functions of these models, and what we expect from a model of the brain. Finally, we argue that a pluralistic approach leading to the integrated coexistence of different models is necessary to study the brain in all its complexity.

Video editing is a high-required job, for it requires skilled artists or workers equipped with plentiful physical strength and multidisciplinary knowledge, such as cinematography, aesthetics. Thus gradually, more and more researches focus on proposing semi-automatical and even fully automatical solutions to reduce workloads. Since those conventional methods are usually designed to follow some simple guidelines, they lack flexibility and capability to learn complex ones. Fortunately, the advances of computer vision and machine learning make up the shortages of traditional approaches and make AI editing feasible. There is no survey to conclude those emerging researches yet. This paper summaries the development history of automatic video editing, and especially the applications of AI in partial and full workflows. We emphasizes video editing and discuss related works from multiple aspects: modality, type of input videos, methology, optimization, dataset, and evaluation metric. Besides, we also summarize the progresses in image editing domain, i.e., style transferring, retargeting, and colorization, and seek for the possibility to transfer those techniques to video domain. Finally, we give a brief conclusion about this survey and explore some open problems.

Early detection improves prognosis in pancreatic ductal adenocarcinoma (PDAC) but is challenging as lesions are often small and poorly defined on contrast-enhanced computed tomography scans (CE-CT). Deep learning can facilitate PDAC diagnosis, however current models still fail to identify small (<2cm) lesions. In this study, state-of-the-art deep learning models were used to develop an automatic framework for PDAC detection, focusing on small lesions. Additionally, the impact of integrating surrounding anatomy was investigated. CE-CT scans from a cohort of 119 pathology-proven PDAC patients and a cohort of 123 patients without PDAC were used to train a nnUnet for automatic lesion detection and segmentation (nnUnet_T). Two additional nnUnets were trained to investigate the impact of anatomy integration: (1) segmenting the pancreas and tumor (nnUnet_TP), (2) segmenting the pancreas, tumor, and multiple surrounding anatomical structures (nnUnet_MS). An external, publicly available test set was used to compare the performance of the three networks. The nnUnet_MS achieved the best performance, with an area under the receiver operating characteristic curve of 0.91 for the whole test set and 0.88 for tumors <2cm, showing that state-of-the-art deep learning can detect small PDAC and benefits from anatomy information.

The rapid infiltration of fake news is a flaw to the otherwise valuable internet, a virtually global network that allows for the simultaneous exchange of information. While a common, and normally effective, approach to such classification tasks is designing a deep learning-based model, the subjectivity behind the writing and production of misleading news invalidates this technique. Deep learning models are unexplainable in nature, making the contextualization of results impossible because it lacks explicit features used in traditional machine learning. This paper emphasizes the need for feature engineering to effectively address this problem: containing the spread of fake news at the source, not after it has become globally prevalent. Insights from extracted features were used to manipulate the text, which was then tested on deep learning models. The original unknown yet substantial impact that the original features had on deep learning models was successfully depicted in this study.

Documenting cultural heritage by using artificial intelligence (AI) is crucial for preserving the memory of the past and a key point for future knowledge. However, modern AI technologies make use of statistical learners that lead to self-empiricist logic, which, unlike human minds, use learned non-symbolic representations. Nevertheless, it seems that it is not the right way to progress in AI. If we want to rely on AI for these tasks, it is essential to understand what lies behind these models. Among the ways to discover AI there are the senses and the intellect. We could consider AI as an intelligence. Intelligence has an essence, but we do not know whether it can be considered “something” or “someone”. Important issues in the analysis of AI concern the structure of symbols -operations with which the intellectual solution is carried out- and the search for strategic reference points, aspiring to create models with human-like intelligence. For many years, humans, seeing language as innate, have carried out symbolic theories. Everything seems to have skipped with the advent of Machine Learning. In this paper, after a long analysis of history, the rule-based and the learning-based vision, we propose KERMIT as a unit of investigation for a possible meeting point between the different learning theories. Finally, we propose a new vision of knowledge in AI models based on a combination of rules, learning and human knowledge.

Foraminifera are single-celled marine organisms that construct shells that remain as fossils in the marine sediments. Classifying and counting these fossils are important in e.g. paleo-oceanographic and -climatological research. However, the identification and counting process has been performed manually since the 1800s and is laborious and time-consuming. In this work, we present a deep learning-based instance segmentation model for classifying, detecting, and segmenting microscopic foraminifera. Our model is based on the Mask R-CNN architecture, using model weight parameters that have learned on the COCO detection dataset. We use a fine-tuning approach to adapt the parameters on a novel object detection dataset of more than 7000 microscopic foraminifera and sediment grains. The model achieves a (COCO-style) average precision of 0.78±0.00 on the classification and detection task, and 0.80±0.00 on the segmentation task. When the model is evaluated without challenging sediment grain images, the average precision for both tasks increases to 0.84±0.00 and 0.86±0.00, respectively. Prediction results are analyzed both quantitatively and qualitatively and discussed. Based on our findings we propose several directions for future work, and conclude that our proposed model is an important step towards automating the identification and counting of microscopic foraminifera.

Accurately mapping individual tree species in densely forested environments is crucial to forest inventory. When considering only RGB images, this is a challenging task for many automatic photogrammetry processes. The main reason for that is the spectral similarity between species in RGB scenes, which can be a hindrance for most automatic methods. State-of-the-art deep learning methods could be capable of identifying tree species with an attractive cost, accuracy, and computational load in RGB images. This paper presents a deep learning-based approach to detect an important multi-use species of palm trees (Mauritia flexuosa; i.e., Buriti) on aerial RGB imagery. In South-America, this palm tree is essential for many indigenous and local communities because of its characteristics. The species is also a valuable indicator of water resources, which comes as a benefit for mapping its location. The method is based on a Convolutional Neural Network (CNN) to identify and geolocate singular tree species in a high-complexity forest environment, and considers the likelihood of every pixel in the image to be recognized as a possible tree by implementing a confidence map feature extraction. This study compares the performance of the proposed method against state-of-the-art object detection networks. For this, images from a dataset composed of 1,394 airborne scenes, where 5,334 palm-trees were manually labeled, were used. The results returned a mean absolute error (MAE) of 0.75 trees and an F1-measure of 86.9%. These results are better than both Faster R-CNN and RetinaNet considering equal experiment conditions. The proposed network provided fast solutions to detect the palm trees, with a delivered image detection of 0.073 seconds and a standard deviation of 0.002 using the GPU. In conclusion, the method presented is efficient to deal with a high-density forest scenario and can accurately map the location of single species like the M flexuosa palm tree and may be useful for future frameworks.

Deep neural networks are hegemonic approaches to many machine learning areas, including natural language processing (NLP). Thanks to the availability of large corpora collections and the capability of deep architectures to shape internal language mechanisms in self-supervised learning processes (also known as "pre-training"), versatile and performing models are released continuously for every new network design. But these networks, somehow, learn a probability distribution of words and relations across the training collection used, inheriting the potential flaws, inconsistencies and biases contained in such a collection. As pre-trained models have found to be very useful approaches to transfer learning, dealing with bias has become a relevant issue in this new scenario. We introduce bias in a formal way and explore how it has been treated in several networks, in terms of detection and correction. Also, available resources are identified and a strategy to deal with bias in deep NLP is proposed.

Rainfall retrieval using geostationary satellites provides critical means to the monitoring of extreme rainfall events. Using the relatively new Himawari 8 meteorological satellite with three times more channels than its predecessors, the deep learning framework of “convolutional autoencoder” (CAE) was applied to the extraction of cloud and precipitation features. The CAE method was incorporated into the Convolution Neural Network version of the PERSIANN precipitation retrieval that uses GOES satellites. By applying the CAE technique with the addition of Residual Blocks and other modifications of deep learning architecture, the presented derivation of PERSIANN operated at the Central Weather Bureau of Taiwan (referred to as PERSIANN-CWB) expands four extra convolution layers to fully use Himawari 8’s infrared and water vapor channels, while preventing degradation of accuracy caused by the deeper network. The development of PERSIANN-CWB was trained over Taiwan for its diverse weather systems and localized rainfall features, and the evaluation reveals an overall improvement from its CNN counterpart and superior performance over all other rainfall retrievals analyzed. Limitation of this model was found in the derivation of typhoon rainfall, an area requiring further research.

The importance of energy systems and its role in economics and politics is not hidden for anyone. This issue is not only important for the advanced industrialized countries, which are major energy consumers, but is also important for oil-rich countries. In addition to the nature of these fuels which contains polluting substances, the issue of their ending up has aggravated the growing concern. Biofuels can be used in different fields for energy production like electricity production, power production or for transportation. Various scenarios have been written about the estimated biofuels from different sources in the future energy system. The availability of biofuels for the electricity market, heating and liquid fuels is very important. Accordingly, the need for handling, modelling, decision making and future forecasting for biofuels can be one of the main challenges for scientists. Recently, machine learning and deep learning techniques have been popular in modeling, optimizing and handling the biodiesel production, consumption and its environmental impacts. The main aim of this study is to evaluate the ML and DL techniques developed for handling biofuels production, consumption and environmental impacts, both for modeling and optimization purposes. This will help for sustainable biofuel production for the future generations.

Wideband signal detection is an important problem in wireless communication. With the rapid development of deep learning (DL) technology, some DL-based methods are applied to wireless technology, and the effect is obvious. In this paper, we propose a novel neural network for multi-type signal detection that can locate signals and recognize signal types in wideband spectrogram. Our network utilizes the key point estimation to locate the rough centerline of signal region and identify class. Then, several regressions are carried out to achieve properties, such as local offset and border offsets of bounding box, which is further synthesized for a more fine location. Experimental results demonstrate that our method performs more accurate than other DL-based object detection methods previously employed for the same detection task. Specifically, our method runs obviously faster than existing methods, and abandons the anchor generation, which makes it more favorable for real-time applications.

The world has been evolving with new technologies and advances day-by-day. With the advent of various learning technologies in every field, the research community is able to provide solution in every aspect of life with the applications of Artificial Intelligence, Machine Learning, Deep Learning, Computer Vision, etc. However, with such high achievements, it is found to lag behind the ability to provide explanation against its prediction. The current situation is such that these modern technologies are able to predict and decide upon various cases more accurately and speedily than a human, but failed to provide an answer when the question of why to trust its prediction is put forward. In order to attain a deeper understanding into this rising trend, we explore a very recent and talked-about novel contribution which provides rich insight on a prediction being made -- ``Explainability.'' The main premise of this survey is to provide an overview for researches explored in the domain and obtain an idea of the current scenario along with the advancements published to-date in this field. This survey is intended to provide a comprehensive background of the broad spectrum of Explainability.

The detection of coronavirus (COVID-19) is now a critical task for the medical practitioner. The coronavirus spread so quickly between people and approaches 100,000 people worldwide. In this consequence, it is very much essential to identify the infected people so that prevention of spread can be taken. In this paper, the deep feature plus support vector machine (SVM) based methodology is suggested for detection of coronavirus infected patient using X-ray images. For classification, SVM is used instead of deep learning based classifier, as the later one need a large dataset for training and validation. The deep features from the fully connected layer of CNN model are extracted and fed to SVM for classification purpose. The SVM classifies the corona affected X-ray images from others. The methodology consists of three categories of Xray images, i.e., COVID-19, pneumonia and normal. The method is beneficial for the medical practitioner to classify among the COVID-19 patient, pneumonia patient and healthy people. SVM is evaluated for detection of COVID-19 using the deep features of different 13 number of CNN models. The SVM produced the best results using the deep feature of ResNet50. The classification model, i.e. ResNet50 plus SVM achieved accuracy, sensitivity, FPR and F1 score of 95.33%,95.33%,2.33% and 95.34% respectively for detection of COVID-19 (ignoring SARS, MERS and ARDS). Again, the highest accuracy achieved by ResNet50 plus SVM is 98.66%. The result is based on the Xray images available in the repository of GitHub and Kaggle. As the data set is in hundreds, the classification based on SVM is more robust compared to the transfer learning approach. Also, a comparison analysis of other traditional classification method is carried out. The traditional methods are local binary patterns (LBP) plus SVM, histogram of oriented gradients (HOG) plus SVM and Gray Level Co-occurrence Matrix (GLCM) plus SVM. In traditional image classification method, LBP plus SVM achieved 93.4% of accuracy.

Restricted Boltzmann machines (RBMs) are the building blocks of some deep learning networks. However, despite their importance, it is our perception that some very important derivations about the RBM are missing in the literature, and a beginner may feel RBM very hard to understand. We provide here these missing derivations. We cover the classic Bernoulli-Bernoulli RBM and the Gaussian-Bernoulli RBM, but leave out the ``continuous'' RBM as it is believed not as mature as the former two. This tutorial can be used as a companion or complement to the famous RBM paper ``Training restricted Boltzmann machines: An introduction'' by Fisher and Igel.

While visual assessment is the standard technique for burn evaluation, computer-aided diagnosis is increasingly sought due to high number of incidences globally. Patients are increasingly facing challenges which are not limited to shortage of experienced clinicians, lack of accessibility to healthcare facilities, and high diagnostic cost. Certain number of studies were proposed in discriminating burn and healthy skin using machine learning leaving a huge and important gap unaddressed; whether burns and related skin injuries can be effectively discriminated using machine learning techniques. Therefore, we specifically use pre-trained deep learning models due to deficient dataset to train a new model from scratch. Experiments were extensively conducted using three state-of-the-art pre-trained deep learning models that includes ResNet50, ResNet101 and ResNet152 for image patterns extraction via two transfer learning strategies: fine-tuning approach where dense and classification layers were modified and trained with features extracted by base layers, and in the second approach support vector machine (SVM) was used to replace top-layers of the pre-trained models, trained using off-the-shelf features from the base layers. Our proposed approach records near perfect classification accuracy of approximately 99.9%.

In recent years, a range of problems under the broad umbrella of computer vision based analysis of ancient coins have been attracting an increasing amount of attention. Notwithstanding this research effort, the results achieved by the state of the art in published literature remain poor and far from sufficiently well performing for any practical purpose. In the present paper we present a series of contributions which we believe will benefit the interested community. We explain that the approach of visual matching of coins, universally adopted in existing published papers on the topic, is not of practical interest because the number of ancient coin types exceeds by far the number of those types which have been imaged, be it in digital form (e.g., online) or otherwise (traditional film, in print, etc.). Rather, we argue that the focus should be on understanding the semantic content of coins. Hence, we describe a novel approach—to first extract semantic concepts from real-world multimodal input and associate them with their corresponding coin images, and then to train a convolutional neural network to learn the appearance of these concepts. On a real-world data set, we demonstrate highly promising results, correctly identifying a range of visual elements on unseen coins with up to 84% accuracy.

Machine learning (ML) has emerged as a critical tool for making sense of the growing amount of genetic and genomic data available because of its ability to find complex patterns in high dimensional and heterogeneous data. While the complexity of ML models is what makes them powerful, it also makes them difficult to interpret. Fortunately, recent efforts to develop approaches that make the inner workings of ML models understandable to humans have improved our ability to make novel biological insights using ML. Here we discuss the importance of interpretable ML, different strategies for interpreting ML models, and examples of how these strategies have been applied. Finally, we identify challenges and promising future directions for interpretable ML in genetics and genomics.

Cognitive neuroscience is the study of how the human brain functions on tasks like decision making, language, perception and reasoning. Deep learning is a class of machine learning algorithms that use neural networks. They are designed to model the responses of neurons in the human brain. Learning can be supervised or unsupervised. Ngram token models are used extensively in language prediction. Ngrams are probabilistic models that are used in predicting the next word or token. They are a statistical model of word sequences or tokens and are called Language Models or Lms. Ngrams are essential in creating language prediction models. We are exploring a broader sandbox ecosystems enabling for AI. Specifically, around Deep learning applications on unstructured content form on the web.

The increasing trend of studying the innate softness of robotic structures and amalgamating it with the benefits of the extensive developments in the field of embodied intelligence has led to sprouting of a relatively new yet extremely rewarding sphere of technology. The fusion of current deep reinforcement algorithms with physical advantages of a soft bio-inspired structure certainly directs us to a fruitful prospect of designing completely self-sufficient agents that are capable of learning from observations collected from their environment to achieve a task they have been assigned. For soft robotics structure possessing countless degrees of freedom, it is often not easy (something not even possible) to formulate mathematical constraints necessary for training a deep reinforcement learning (DRL) agent for the task in hand, hence, we resolve to imitation learning techniques due to ease of manually performing such tasks like manipulation that could be comfortably mimicked by our agent. Deploying current imitation learning algorithms on soft robotic systems have been observed to provide satisfactory results but there are still challenges in doing so. This review article thus posits an overview of various such algorithms along with instances of them being applied to real world scenarios and yielding state-of-the-art results followed by brief descriptions on various pristine branches of DRL research that may be centers of future research in this field of interest.

We examine the use of deep learning (neural networks) to predict the movement of the S&P 500 Index using past returns of all the stocks in the index. Our analysis finds that the future direction of the S&P 500 index can be weakly predicted by the prior movements of the underlying stocks in the index. Decomposition of the prediction error indicates that most of the lack of predictability comes from randomness and only a little from nonstationarity. We believe this is the first test of S&P500 market efficiency that uses a very large information set, and it extends the domain of weak-form market efficiency tests.

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.

An adaptive deep neural network is used in an inverse system identification setting to approximate the inverse of a nonlinear plant with the aim of constituting the plant controller by copying to the latter the weights and architecture of the converging deep neural network. This deep learning (DL) approach to the adaptive inverse control (AIC) problem is shown to outperform adaptive filtering techniques and algorithms normally used in adaptive control, especially when the plant is nonlinear. The deeper the controller the better the inverse function approximation, provided that the nonlinear plant have an inverse and that this inverse can be approximated. Simulation results prove the feasibility of this DL-based adaptive inverse control scheme. The DL-based AIC system is robust to parameter change of the nonlinear plant in that, under such change, the plant output reassumes the value of the reference signal considerably faster than with the adaptive filter counterpart of the deep neural network. Settling times and rise times of the step response are shown to improve in the DL-based AIC system.

Abstract: The main contribution of this paper is to introduce a new iterative training algorithm for restricted Boltzmann machines. The proposed learning path is inspired from online sequential extreme learning machine one of extreme learning machine variants which deals with time accumulated sequences of data with fixed or varied sizes. Recursive least squares rules are integrated for weights adaptation to avoid learning rate tuning and local minimum issues. The proposed approach is compared to one of the well known training algorithms for Boltzmann machines named “contrastive divergence”, in term of time, accuracy and algorithmic complexity under the same conditions. Results strongly encourage the new given rules during data reconstruction.

A machine learning (ML)-based methodology for predicting frosts is applied to the southern and southeastern regions of Brazil, as well as to other countries including Uruguay, Paraguay, northern Argentina, and southeastern Bolivia. The machine learning model (using TensorFlow (TF)) was compared to the frost index IG (from the Portuguese: Índice de Geada) developed by the National Institute for Space Research (INPE, Brazil). The IG index is estimated using meteorological variables from a regional weather numerical model (RWNM). After calculating the two indices using the ML model and the RWNM, a voting committee (VC) was trained to select between the computed outputs. The AdaBoostClassifier algorithm was employed to implement the voting committee. The study area was subdivided into three distinct subregions: R1 (outside Brazil), R2 (the southern of Brazil), and R3 (the southeastern of Brazil). Two forecasting time scales were evaluated: 24 hours and 72 hours. The 24-hour forecasts from both approaches (TF and RWNM) exhibited similar performance in terms of the number of accurate predictions. However, in the region covering Uruguay and northern Argentina, the TensorFlow model demonstrated superior frost prediction accuracy. Additionally, the TensorFlow model outperformed the RWNM for the 72-hour forecast horizon.

Predicting stock and cryptocurrency prices is challenging. One method involves ana-lyzing opinions from individuals and experts. This article focuses on analyzing social media sen-timents for predicting Bitcoin prices using various criteria. In this article, unlike other articles that only used the points obtained from sentiment analysis of tweets, we combined the points of tweets based on other criteria, such as the highest like counts, the highest volume of tweets, and the weighted scores of tweets. In the first stage, we evaluated the deep learning models SL, LG, LGC, BL, and BG without considering sentiment. Ultimately, the BG model performed well in evaluation metrics such as MAE and Loss. In the second stage, we examined the impact of tweet volume on the models' performance. In the third stage, we utilized the simulation results of the VADER model. For the fourth stage, we considered the maximum likes for tweets as a sentiment analysis tool. Finally, in the last stage, which is the main focus of this article, we employed weighted averaging for sentiment analysis. Model BG provided approximately 11% improvement based on the MAE metric and about 15% improvement based on the Loss metric for better detection the following day.

In this paper we present a novel auto-probabilistic mining method designed to enhance the training process of Siamese neural networks. The proposed method demonstrates significant efficiency, particularly in scenarios characterized by limited data and computational resource. The study incorporates the application of the newly developed mining method in conjunction with a previously introduced auto-clustering technique. Comparative experiments are conducted both with and without the integration of the proposed method, highlighting its impact on overall performance. Additionally, the research introduces a novel metric loss to further refine the training process taking into account the drawbacks of existing loss functions applying for metric learning. During the experimentation phase, validation is conducted on the printed Hangul character PHD08 database and Omniglot. This validation provides valuable insights into the performance and effectiveness of the proposed methods.

Early detection of cervical cancer is the need of the hour to stop the fatality from this disease. There have been various CAD approaches in the past that promise to detect cervical cancer in an early stage, each of them having some constraints.This study proposes a novel ensemble deep learning model, VLR (variable learning rate), aimed at enhancing the accuracy of cervical cancer classification. The model architecture integrates VGG16, Logistic Regression, and ResNet50, combining their strengths in an offbeat ensemble design. VLR learns in two ways: firstly, by dynamic weights from three base models, each of them trained separately; secondly, by attention mechanisms used in the dense layer of the base models. Hyperparameter tuning is applied to further reduce loss, fine tune the model’s performance, and maximize classification accuracy. We performed K-fold cross-validation on VLR to evaluate any improvements in metric values resulting from hyperparameter fine tuning. We have also validated our model on images captured in three different solutions and on a secondary dataset. Our proposed VLR model outperformed existing methods in cervical cancer classification, achieving a remarkable training accuracy of 99.95% and a testing accuracy of 99.89%.

Wideband signal detection and identification are crucial for military and civilian fields such as communication reconnaissance and spectrum security. This paper systematically sorts out the mainstream technologies of wideband signal detection and identification, and expounds the main current short-wave wideband signal detection and identification work from three perspectives: traditional expert feature methods, combination of expert features and machine learning, and wideband signal detection and identification based on deep learning. Particularly, this paper focuses on the processing methods of current key signals such as short burst signals and frequency hopping signals, as well as the latest deep learning methods in the past five years. Finally, the shortcomings of the existing broadband signal detection and identification methods are summarized, and the future trend is predicted.

Background Anemia is a prevalent global health issue, affecting individuals across all age groups and leading to significant morbidity. Traditional methods for diagnosing anemia rely on laboratory assessments of hemoglobin concentration, which are invasive and costly. This study investigates the potential of a chest X-ray (CXR)-based deep learning model (DLM) to predict anemia, offering a non-invasive and efficient diagnostic alternative. Patients and methods We conducted a retrospective cohort study using data from Tri-Service General Hospital, Taipei, Taiwan, collected between June 2016 and February 2022. A total of 305,793 patients aged 20 years and older who had at least one CXR were included. The dataset was divided into development, tuning, internal validation, and external validation sets. All data originated from Tri-Service General Hospital, except for the external validation set, which was obtained from the Jingjhou branch. A DLM based on a 50-layer SE-ResNeXt architecture was developed to predict anemia from CXR images. The model's performance was compared to predictions made using traditional clinical data to assess its predictive accuracy. Results The DLM for predicting anemia showed strong performance, with an AUC of 0.845 in the internal validation set (sensitivity 68.5%, specificity 84.7%) and 0.852 in the external set (sensitivity 71.5%, specificity 83.2%). Subgroup analysis revealed better diagnostic accuracy with computed radiography, posteroanterior (PA) view, and in males aged 55 to 64, while younger females had lower accuracy. The DLM, especially when combined with patient data, outperformed other clinical models. Conclusion Our study introduces a deep learning tool for detecting anemia from chest images, offering faster decision-making and interventions in clinical settings. With further refinement, this model aims to align more closely with real-time clinical conditions, enhancing its utility. Ultimately, it has the potential to improve patient care through quicker anemia detection and better clinical outcomes.

Healthcare is undergoing a technological transformation, particularly in robotic surgery and arti-ficial intelligence (AI) applied to cytology. These innovations are improving the precision and effi-ciency of both surgical procedures and laboratory diagnostics. Robotic systems, such as the Da Vinci, enable surgeons to perform complex procedures with enhanced control and precision, leading to minimally invasive surgeries with reduced recovery times. This technology has signifi-cantly impacted urological, cardiac, and gastrointestinal surgeries, and its use is expected to grow. In cytology, AI automates the analysis of cell samples, leading to faster and more accurate diag-noses, especially in detecting diseases like cervical cancer. AI reduces errors and enhances diag-nostic accuracy by identifying complex patterns in cellular images. Predictive analytics, driven by big data, also allows for personalized treatments based on patient-specific data. However, these technologies face challenges, including high costs, ethical concerns, data privacy issues, and resistance from healthcare professionals. Continuous training is essential for medical personnel to effectively adopt these tools. Furthermore, the rise of telemedicine and telesurgery, supported by advancements in 5G, offers new opportunities for remote care but also introduces risks, such as depersonalization of medical care and technological barriers. Ultimately, while robotic surgery and AI hold great promise for improving clinical outcomes, their successful integration will depend on addressing these challenges and ensuring that healthcare professionals are adequately prepared for the digital future.

Diabetic retinopathy, a common complication of diabetes, is further exacerbated by factors such as hypertension and obesity. This study introduces the Diabetic Retinopathy Convolutional Transformer (DRCT) model, which combines convolutional and transformer techniques to enhance the classification of retinal images. The DRCT model achieved an impressive average F1 score of 0.97, reflecting its high accuracy in detecting true positives while minimizing false positives. Throughout 100 training epochs, the model exhibited strong generalization capabilities, achieving superior validation accuracy with minimal overfitting. On a newly evaluated dataset, the model attained precision and recall scores of 96.93% and 98.89%, respectively, indicating a well-balanced handling of false positives and false negatives. The model's ability to classify retinal images into five distinct diabetic retinopathy categories demonstrates its potential to significantly improve automated diagnosis and aid in clinical decision-making.

Automating pruning tasks entails overcoming several challenges, encompassing not only robotic manipulation but also environment perception and detection. To achieve efficient pruning, robotic systems must accurately identify the correct cutting points. A possible method to define these points is to choose the cutting location based on the number of nodes present on the targeted cane. For this purpose, in grapevine pruning, it is required to correctly identify the nodes present on the primary canes of the grapevines. In this paper, a novel method of node detection in grapevines is proposed with four distinct state-of-the-art versions of the YOLO detection model: YOLOv7, YOLOv8, YOLOv9 and YOLOv10. These models were trained on a public dataset with images containing artificial backgrounds and afterwards validated on different cultivars of grapevines from two distinct Portuguese viticulture regions with cluttered backgrounds. This allowed to evaluate the robustness of the algorithms on the detection of nodes in diverse environments, compare the performance of the YOLO models used, as well as create a publicly available dataset of grapevines obtained in Portuguese vineyards for node detection. Overall, all used models were capable of achieving correct node detection on images of grapevines from the three distinct datasets. Considering the trade-off between accuracy and inference speed, the YOLOv7 model demonstrated to be the most robust in detecting nodes in 2D images of grapevines, achieving F1-Score values between 70 % and 86.5 % with inference times of around 89 ms for an input size of 1280×1280 px. Considering these results, this work contributes with an efficient approach for real-time node detection for further implementation on an autonomous robotic pruning system.

Traditional classroom teaching model can no longer meet the needs of students' ability development. As a new model actively advocated in the modern education concept, personalized teaching focuses on the development needs of students and emphasizes respect for students' individualized growth. In this work, we utilize deep learning method to recommend teaching strategies for different features of students. Specifically, the proposed method includes the construction of a multi-layer neural network model, which can analyze students' multi-dimensional learning data and behavior patterns, so as to recommend the most suitable teaching strategies for individual students. Through data collection and preprocessing, model architecture design, model training and optimization, the model gradually improves the accuracy of prediction and recommendation. Experimental analysis verifies the effectiveness of the model through field teaching experiments, and the results show that the personalized teaching strategy based on deep learning can significantly improve the learning effect and participation of students.

The continuous expansion of Large Language Models (LLMs) has significantly transformed the fields of artificial intelligence (AI) and natural language processing (NLP). This paper reviews the rapidly evolving domain of language models and introduces the concept of Extremely Large Language Models (XtremeLLMs), a new category defined for models exceeding one trillion parameters. These models are monumental in scale and engineered to enhance performance across a diverse range of language tasks. This study aims to establish a comprehensive framework that explores the significant opportunities and complex challenges presented by such extensive scaling and emphasises the implications for future advancements in the field.

Accurate assessment of shale gas reserves hinges on a detailed characterization of gas and water distribution within shale reservoirs. Despite its importance, the nanoscale behavior of water and methane in deep shale formations, especially under elevated temperatures and pressures, remains poorly understood. This study proposes a novel molecular modeling approach to represent the complex nanocomposite porous media found in deep shale matrices. Leveraging quantitative data from experimental analyses of shale pore composition, we developed a molecular model that accurately reflects the intricate pore-slit structure within a composite of kerogen and illite, common constituents of shale. Employing a hybrid grand canonical Monte Carlo and molecular dynamics simulation technique, we investigated the micro-scale distribution and preferential adsorption sites of water and methane within the shale matrix. Our research further elucidates the impact of water on the pore structure of shale and delineates the co-occurrence patterns of water and methane. The simulations reveal that water and methane share similar adsorption preferences on kerogen surfaces, while methane does not exhibit specific adsorption preferences on water-bearing illite. We also find that the presence of water significantly alters the illite pore space, leading to a reduction in larger pore volumes and a concomitant increase in smaller pores. The decrease in methane content with increasing water content is more pronounced in the shrinkage cracks compared to the kerogen matrix pores and is particularly notable in the macropores of illite. The findings underscore the differential sensitivity of methane content to water levels across various shale pore types. These insights contribute to a deeper theoretical understanding of gas and water distribution mechanisms in deep shale reservoirs, offering valuable support for the evaluation of deep shale gas resource potential.

Chip pad alignment inspection is of great importance in the industrial field. However, due to the fact that chip pads are usually small, problems such as misdetection and missed detection often occur. When applying deep learning methods for chip pad detection, it is necessary to ensure accurate detection of small target chips while meeting the requirements of lightweight detection models for industrial needs. To solve the above problems, this paper proposes a lightweight model based on improved YOLOv5s. Firstly, the feature extraction part is improved to increase the network's focus on the target. Secondly, the feature fusion layer is improved to double the resolution of the prediction head, and the context-aware network is designed to enhance the context-capture ability of key features of small targets. Finally, SIoU is adopted as the loss function to improve the speed and accuracy of the regression frame. The experimental results show that the improved YOLOv5s algorithm improves the detection accuracy by 2.3% and reduces the network parameters by 81.8% compared to the original algorithm. The improved algorithm is combined with image processing techniques to design correction methods for alignment anomalies and realize real-time alignment anomaly correction in industry.

Recurrent Neural Networks (RNNs) have significantly advanced the field of machine learning by enabling the effective processing of sequential data. This paper provides a comprehensive review of RNNs and their applications, highlighting advancements in architectures such as Long Short-Term Memory (LSTM) networks, Gated Recurrent Units (GRUs), Bidirectional LSTM (BiLSTM), and stacked LSTM. The study examines the application of RNNs in different domains, including natural language processing (NLP), speech recognition, financial time series forecasting, bioinformatics, autonomous vehicles, and anomaly detection. Additionally, the study discusses recent innovations, such as the integration of attention mechanisms and the development of hybrid models that combine RNNs with convolutional neural networks (CNNs) and transformer architectures. This review aims to provide machine learning researchers and practitioners with a comprehensive overview of the current state and future directions of RNN research.

Attention Deficit Hyperactivity Disorder (ADHD) diagnosis poses significant challenges due to its heterogeneous nature and reliance on subjective assessments. This study leverages deep learning to develop a robust diagnostic model by integrating a comprehensive set of EEG biomarkers. We incorporate parameters such as theta/beta ratio, coherency measures, delta power, event-related potentials (ERPs), power spectral density (PSD), microstates, entropy measures, fractal dimension, and source localization. Our convolutional neural network (CNN) model, designed to process these diverse features, demonstrates high accuracy and stability in distinguishing ADHD patients from controls, from all ages. The model's architecture includes convolutional layers for spatial feature extraction, followed by dense layers that integrate additional EEG parameters. The results indicate that deep learning, coupled with a rich feature set, can significantly enhance the predictive accuracy of ADHD diagnosis, offering a promising tool for clinical application.

Autoencoders are neural networks that have applications in denoising processes. Their use is widely reported in imaging (2D), though 1D series can also benefit from this function. Here, three canonical waveforms are used to train a neural network and achieve a signal-to-noise reduction with curves whose noise energy was above that of the signals. A real-world test is carried out with the same autoencoder subjected to a set of time series corrupted by noise generated by a zener diode, biased on the avalanche region. Results showed that, observed some guidelines, the autoencoder can indeed denoise 1D waveforms usually observed in electronics, particularly square waves found in digital circuits.

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.

With the rapid developments that have been made in computer vision and computer technology, semantic segmentation of high-resolution images has emerged as a mainstream and challenging task. The, segmentation of water areas remains a complex task due to the similar features between water areas or between water areas and other land objects. Meanwhile, since coastlines and riverbanks are typically presented in irregular shapes, it is difficult to obtain the categories of pixels at boundaries. This paper proposed the use of a Water-Land Boundary Attention Network (WLBANet) to improve the accuracy of pixels at boundaries between water and land. This network also reinforced the performance of contextual information extraction into distinguishing between water areas and other land objects. To prove the validity of this proposed network, WLBANet was performed on the Water Land Dataset (WLD), which is also by this paper, and which contains water areas and other land objects categories. The results demonstrates that the proposed method has significant effectiveness.

Hyperspectral images (HSI) contain abundant spectral information. Efficient extraction and utilization of this information for image classification remain prominent research topics. Previously, hyperspectral classification techniques primarily relied on statistical attributes and mathematical models of spectral data. Deep learning classification techniques have recently been extensively utilized for hyperspectral data classification, yielding promising outcomes. This study proposes a deep learning approach that uses polarization feature maps for classification. Initially, the polar coordinate transformation method is employed to convert the spectral information of all pixels in the image into spectral feature maps. Subsequently, the proposed Deep Context Feature Fusion Network (DCFF-NET) is utilized to classify these feature maps. The model is validated using three open-source hyperspectral datasets: Indian Pines, Pavia University, and Salinas. The experimental results indicated that DCFF-NET achieves excellent classification performance. Experimental results on three public HSI datasets demonstrate that the proposed method accurately recognizes different objects with an overall accuracy (OA) of 86.68%, 94.73%, and 95.14% based on the pixel method, and 98.15%, 99.86%, and 99.98% based on the pixel-patch method.

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.

With increasing electronic medical data and the development of artificial intelligence, Clinical Decision Support Systems (CDSSs) assist clinicians in diagnosis and prescription. Traditional knowledge-based CDSSs follow an accumulated medical knowledgebase and a predefined rule system, which clarifies the decision-making process; however, maintenance cost issues exist in the medical data quality control and standardization process. Non-knowledge-based CDSSs utilize vast amounts of data and algorithms to effectively decide; however, the deep learning black-box problem causes unreliable results. EXplainable Artificial Intelligence (XAI)-based CDSS provides a valid rationale and explainable results. It ensures trustworthiness and transparency by showing the recommendation and prediction results process through explainable techniques. However, existing systems have limitations, such as the scope of data utilization and the lack of explanatory power of AI models. This study proposes a new XAI-based CDSS framework to address these issues; introduce resources, datasets, and models that can be utilized; and provides a foundation model to support decision-making in various disease domains. Finally, we propose future directions for CDSS technology and highlight societal issues that need to be addressed to emphasize the potential of CDSS in the future.

Fluorescence microscopic images of cells contain a large number of morphological features that are used as an unbiased source of quantitative information about cell status, through which researchers can extract quantitative information about cells and study the biological phenomena of cells through statistical and analytical analysis. As an important research object of phenotypic analysis, images have a great influence on the research results. Saturation artifacts present in the image result in a loss of grayscale information that does not reveal the true value of fluorescence intensity. From the perspective of data post-processing, we propose a two-stage cell image recovery model based on generative adversarial network to solve the problem of phenotypic feature loss caused by saturation artifacts. The model is capable of restoring large areas of missing phenotypic features. In the experiment, we adopt the strategy of progressive restore to improve the robustness of the training effect, and add the contextual attention structure to enhance the stability of the restore effect. We hope to use deep learning methods to mitigate the effects of saturation artifacts to reveal how chemical, genetic, and environmental factors affect cell state, providing an effective tool for studying the field of biological variability and improving image quality in analysis.

In the intersection of the Internet of Things (IoT) and Machine Learning (ML), the choice between high-level and low-level programming libraries presents a significant dilemma for developers, impacting not only the efficiency and effectiveness of ML models but also their environmental footprint. We have proposed a comprehensive framework to aid in this decision-making process, underpinned by a detailed comparative analysis of both types of libraries on one of the key IoT ML task; image classification. We have introduced a novel algorithm designed to calculate the green footprint of ML model training, factoring in execution time, memory utilization, power consumption, and CPU temperature, addressing the urgent need for sustainable ML practices. Through an empirical evaluation of popular libraries such as PyTorch for high-level and Libtorch for low-level development, we have assessed their performance, development efficiency, and hardware compatibility. The culmination of our research is a decision support system that synthesizes the experimental findings to guide developers toward choices that harmonize model performance with environmental sustainability.

Deep learning methods paired with sidescan sonar (SSS) are commonly used in underwater search and rescue operations for drowning victims, wrecks, and airplanes. However, these techniques are primarily designed to detect mine-like objects and are rarely applied to identifying features in dynamic dredge pit environments, due to a lack of datasets. In this study, we present a Sandy Point Dredge Pit (SPDP) dataset, in which high-resolution SSS data were collected from the west flank of Mississippi bird foot delta of the Louisiana inner shelf. This dataset contains a total of 385 SSS images. We introduce an Effective Geomorphology Classification model (EGC). By ablation studies, we analyze the utility of transfer learning on different model architectures and the impact of data augmentations on model performance. This EGC model will make geomorphic features identification in dredge pit environments quick and efficient which requires extensive experience and professional knowledge. The combination of SSS images and EGC model is a cost-effective and valuable toolkit for hazards monitoring in marine dredge pit environment. The SPDP SSS images dataset, especially the feature of pit wall without rotational slump, is also valuable for other machine learning modelers.

High-power laser facilities necessitate predicting incremental damage to final optics to identify evolving damage trends. In this study, we propose a surface damage detection method utilizing image segmentation employing ResNet-18, and a damage area estimation network employing U-Net++. Paired sets of online and offline images of optics obtained from a large laser facility is used to train the network. The trends of varying damage could be identified by incorporating additional experimental parameters. A key advantage of the proposed method is that the network can be trained end-to-end on small samples, eliminating the need for manual labeling or feature extraction. The software developed based on these models can facilitate the daily inspection and maintenance of optics in large laser facilities.

The Ebola virus disease (EVD) is an extremely contagious and fatal illness caused by the Ebola virus. Recently, Uganda witnessed an outbreak of EVD, which generated a lot of attention on various social media platforms. To ensure effective communication and implementation of targeted health interventions, it is crucial for stakeholders to comprehend the sentiments expressed in the posts and discussions on these online platforms. In this study, we used deep learning techniques to analyze the sentiments expressed in Ebola-related tweets during the outbreak. We explored the application of three deep learning techniques to classify the sentiments in 8,395 tweets as positive, neutral, or negative. The techniques examined included a 6-layer convolutional neural network (CNN), a 6-layer long-short-term memory model (LSTM), and an 8-layer Bidirectional Encoder Representations from Transformers (BERT) model. The study found that the BERT model outperformed both the CNN and LSTM-based models across all the evaluation metrics, achieving a remarkable classification accuracy of 95\%. These findings confirm the reported effectiveness of transformer-based architectures in tasks related to natural language processing, such as sentiment analysis.

Despite significant advances in tumor biology and clinical therapeutics, metastasis remains the primary cause of cancer-related deaths. While RNA-seq technology has been used extensively to study metastatic cancer characteristics, challenges persist in acquiring adequate transcriptomic data. To overcome this challenge, we propose MetGen, a generative contrastive learning based on deep learning model. MetGen generates synthetic metastatic cancer expression profiles using primary cancer and normal tissue expression data. Our results demonstrate that MetGen generates comparable samples to actual metastatic cancer samples, and we discuss the learning mechanism of MetGen. Additionally, we demonstrate MetGen's interpretability using metastatic prostate cancer and metastatic breast cancer. MetGen has learned highly relevant signatures in cancer, tissue, and tumor microenvironment, such as immune response and the metastasis process, which potentially fosters a more comprehensive understanding of metastatic cancer biology. The development of MetGen represents a significant step toward the study of metastatic cancer biology by providing a generative model that identifies candidate therapeutic targets for the treatment of metastatic cancer.

The incessant evolution of online platforms has ushered in a multitude of shopping modalities. Within the food industry, however, assessing the delectability of meals can only be tentatively determined based on consumer feedback encompassing aspects such as taste, pricing, packaging, service quality, delivery timeliness, hygiene standards, and environmental considerations. Traditional text data mining techniques primarily focus on consumers' emotional traits, disregarding pertinent information pertaining to the online products themselves. In light of these aforementioned issues in current research methodologies, this paper introduces the Bert BiGRU Softmax model combined with multimodal features to enhance the efficacy of sentiment classification in data analysis. Comparative experiments conducted using existing data demonstrate that the accuracy rate of the model employed in this study reaches 90.9%. In comparison to single models or combinations of three models with the highest accuracy rate of 7.7%, the proposed model exhibits superior accuracy and proves highly applicable to online reviews.

In the era of Artificial Intelligence (AI), comprehending and responding to non-verbal communication is increasingly vital. This research extends AI's reach in bridging communication gaps, notably benefiting American Sign Language (ASL) and Taiwan Sign Language (TSL) communities. It focuses on employing various AI models, especially the Hierarchical Vision Transformer with Shifted Windows (Swin), for recognizing diverse sign language datasets. The study assesses Swin architecture's adaptability to different sign languages, aiming to create a universal platform for Unvoiced communities. Utilizing deep learning and transformer technologies, hybrid application prototypes have been developed for ASL-to-English translations and vice versa, with plans to expand this to multiple sign languages. The Swin models, trained on varied dataset sizes, show considerable accuracy, indicating their flexibility and effectiveness. This research underscores major advancements in sign language recognition and underlines a commitment to inclusive communication in the digital era. Future work will focus on enhancing these models and broadening their scope to include more sign languages, integrating multimodality and Large Language Models (LLMs), thereby fostering global inclusivity.

Background: Silent MRA has shown promising results in evaluating the stents used for intracranial aneurysm treatment. A deep learning-based algorithm was recently introduced by GE. The purpose of this study was to compare the performance of several MRA techniques in lumen visibility in silicone models with flow diverter stents. Methods: Two Surpass Evolve stents of different sizes were implanted in two silicone tubes. The tubes were placed in separate boxes in straight position and in two different curve configurations and connected to a pulsatile pump to construct a flow loop. Using a GE 3.0T scanner, TOF and silent MRA images were acquired, and deep learning reconstruction was applied to the silent MRA dataset. Intra-luminal signal intensity in the stent (SIin-stent), in the tube outside the stent (SIvessel) and of the background (SIbg) were measured for each scan. Results: The SIin-stent/SIbg and SIin-stent/SIv ratio were higher on silent scans and DL-based reconstructions than on TOF images. The stent tips created severe artefacts on TOF images, which could not be observed on silent scans. Conclusions: Our study demonstrated that the DL algorithm improved the quality of the silent technique in evaluating the flow diverter stent patency.

Document image binarization is a challenging task, especially when it comes to text segmentation in degraded document images. The binarization, as a pre-processing step of optical character recognition (OCR), is one of the most fundamental and commonly used segmentation methods. It separates the foreground text from the background of the document image to facilitate subsequent image processing. In view of the different degradation degree of document image, researchers have proposed a variety of solutions. This paper reviews the main binarization techniques, including both traditional algorithms and deep learning-based algorithms. We also summarize some difficulties and challenges in the field of document image binarization. Here, we evaluate various image binarization techniques to identify shortcomings in current methods and provide some help for future research.

Vulnerabilities in smart contracts may trigger serious security events, and the detection of smart contract vulnerabilities has become a significant problem. In this paper, by using the multi-scale cascade encoder architecture as the backbone, we propose a novel Multi-scale Encoder Vulnerability Detection (MEVD) approach to detect well-known high-risk vulnerabilities in smart contracts. Firstly, we use the gating mechanism to design a unique Surface Feature Encoder (SFE) to enrich the semantic information of code features. Then, by combining a Base Transformer Encoder (BTE) and a Detail CNN Encoder (DCE), we introduce a dual-branch encoder to capture the global structure and local detail features of the smart contract code, respectively. Finally, to focus the model's attention on vulnerability-related characteristics, we employ the Deep Residual Shrinkage Network (DRSN). Experimental results on three types of high-risk vulnerability datasets demonstrate performance compared to state-of-the-art methods, and our method achieves an average detection accuracy of 90%.

In recent years, Edge Intelligence (EI) has emerged, combining edge computing with AI, specifically deep learning, to run AI algorithms directly on edge devices. In practical applications, EI faces challenges related to computational power, power consumption, size, and cost, with the primary challenge being the trade-off between computational power and power consumption. This has rendered traditional computing platforms unsustainable, making heterogeneous parallel computing platforms a crucial pathway for implementing EI. In our research, we leveraged the Xilinx Zynq 7000 heterogeneous computing platform, employed High-Level Synthesis (HLS) for design, and implemented two different accelerators for LeNet-5 using loop unrolling and pipelining optimization techniques. The experimental results show that when running at a clock speed of 100 MHz, the PIPELINE accelerator, compared to the UNROLL accelerator, experiences an 8.09% increase in power consumption but speeds up by 14.972 times, making the PIPELINE accelerator superior in performance. Compared to the CPU, the PIPELINE accelerator reduces power consumption by 91.37% and speeds up by 70.387 times, while compared to the GPU, it reduces power consumption by 93.35%.This study provides two different optimization schemes for edge intelligence applications through design and experimentation, and demonstrates the impact of different quantization methods on FPGA resource consumption. These experimental results can provide a reference for practical applications, thereby providing a reference hardware acceleration scheme for edge intelligence applications.

Today, specific CNN models assigned to specific tasks are often used. In this article, the authors have investigated three models MobileNet, EfficientNetB0 and InceptionV3. Artificial intelligence methods for deep networks will be presented but implemented in a limited computer resource. Three types of training bases were investigated, starting with a simple base verifying five colors, followed by recognition of two different orthogonal elements, and then more complex images discriminating between elements. The research aimed to demonstrate the models' capabilities based on training base parameters such as the number of images and epoch types. The architectures proposed by the authors in these cases were chosen based on simulation studies conducted on a virtual machine with limited hardware parameters. The proposals present the advantages and disadvantages of different models based on TensorFlow, Keras libraries in the Jupiter environment written based on the Python language. The authors' proposal of the selected AI model enables further work on e.g. image classification, but in a limited computer resource for industrial implementation e.g. on a PLC.

This paper presents an approach to enhancing the clarity and intelligibility of speech in digital communications compromised by various background noises. Utilizing deep learning techniques, specifically a Variational Autoencoder (VAE) with 2D convolutional filters, we aim to suppress background noise in audio signals. Our method focuses on four simulated environmental noise scenarios: storms, wind, traffic, and aircraft. Training dataset has been obtained from public sources (TED-LIUM 3 dataset, which includes audio recordings from the popular TED-TALK series) combining with these background noises. The audio signals were transformed into 2D power spectrograms, upon which our VAE model was trained to filter out the noise and reconstruct clean audio. Our results demonstrate that the model outperforms existing state-of-the-art solutions in noise sup-pression. Although differences in noise types were observed, it was challenging to definitively conclude which background noise most adversely affects speech quality. Results have been assessed with objective methods (mathematical metrics) and subjective (listening to a set of audios by humans). Notably, wind noise showed the smallest deviation between the noisy and cleaned audio, perceived subjectively as the most improved scenario. Future work involves refining the phase calculation of the cleaned audio and creating a more balanced dataset to minimize differences in audio quality across scenarios. Additionally, practical ap-plications of the model in real-time streaming audio are envisaged. This research contributes significantly to the field of audio signal processing by offering a deep learning solution tailored to various noise conditions, enhancing digital communication quality.

Measuring Marginal Reflex Distance-1 (MRD-1) is a crucial clinical tool used to evaluate the position of the eyelid margin in relation to the cornea. Traditionally, this assessment has been conducted manually by plastic surgeons, ophthalmologists, or trained technicians. However, with the advancements in Artificial Intelligence (AI) technology, there is a growing interest in the development of automated systems capable of accurately measuring MRD-1.In this context, we introduce novel MRD-1 measurement methods based on deep learning algorithms that can simultaneously capture images and compute the results. This prospective observational study involved 154 eyes of 77 patients aged over 18 years who visited Chungnam National University Hospital between September 1, 2023, and July 29, 2023. We collected four different MRD1 da-tasets from patients using three distinct measurement methods, each tailored to the individual patient. The mean MRD1 values, measured through the manual method using a penlight, the deep learning method, ImageJ analysis from RGB eye images, and ImageJ analysis from IR eye images in 56 eyes of 28 patients, were 2.64 ± 1.04 mm, 2.85 ± 1.07 mm, 2.78 ± 1.08 mm, and 3.07 ± 0.95 mm, respectively. Notably, the strongest agreement was observed between MRD1_deep learning (DL) and MRD1_IR (0.822, p < 0.01). In the Bland-Altman plot, the smallest difference was observed between MRD1_DL and MRD1_IR ImageJ, with a mean difference of 0.0611 and △LOA (limits of agreement) of 2.5162, which is the smallest among the other groups. In conclusion, this novel MRD1 measurement method, based on an IR camera and deep learning, demon-strates statistical significance and can be readily applied in clinical settings.

With the rapid development of space programs in various countries, the number of satellites in space is increasing, resulting in an increasingly complex space environment. Therefore, improving space object identification technology has become highly important. We proposes a method of applying deep learning to intelligent detection of space object. We utilize 49 authentic 3D satellite models including 16 scenarios to generate a dataset comprising 17,942 images, which contains over 500 actual satellite photos. Additionally, we acquired a substantial amount of annotated data using a semi-automatic labeling method, which resulted in significant labor cost savings, and obtained a total of 39,000 labels. We validate the feasibility of the dataset using YOLOv3 and YOLOv7 models. What's more, we optimize the YOLOv7 model by integrating deformable convolution RepPoint into the YOLOv7 backbone to obtain the YOLOv7-R model. Through training with these two models, experimental results show that YOLOv3 achieves an accuracy of 0.927, YOLOv7 reaches an accuracy of 0.964, and YOLOv7-R achieves the highest accuracy at 0.983. This provides an effective solution for intelligent space object detection.

In the realm of autonomous vehicle environment perception, the primary objective of point cloud target detection is to swiftly and accurately identify three-dimensional objects from point cloud data. To meet this requirement, the prevalent network architecture employed in the industry is the voxel-based PointPillars model. Nonetheless, this model faces challenges in maintaining detection accuracy when objects are obscured or diminutive in size. In response to this issue, we introduce AGPNet, a novel model that seamlessly integrates four key modules: Data Augmentation, Dynamic Graph CNN, Pillar Feature Net, and Detection Head (SSD). The Data Augmentation module enhances the adaptability of point cloud data to complex and ever-changing real-world environments. The Dynamic Graph CNN module endows the network structure with geometric features, which encapsulate not only the point itself but also its adjacent points. The Pillar Feature Net module translates three-dimensional point cloud data into pseudo-image data through the utilization of voxels. Subsequently, the Detection Head (SSD) module leverages this pseudo-image data to conduct target detection of three-dimensional objects. Our experiments, conducted on the KITTI dataset, demonstrate that our proposed method boosts object detection accuracy by 6-7 percentage points compared to the PointPillars model, while maintaining similar detection times.

In this paper, we introduce an innovative approach to distinguish Gliosarcoma (GSM) from Glioblastoma (GBM). Our method combines causal fuzzy logic rules with the Big Bird architecture, a Transformer-based Deep Learning algorithm. Unlike prior research, which often relied on statistical models to reduce dataset dimensions before causal analysis, our approach harnesses the complete dataset in tandem with our causal fuzzy Big Bird architecture. Additionally, we benchmark our results not only against previous Gliosarcoma/Glioblastoma studies but also against GPT-2 for a comprehensive evaluation.

CagY is the largest and most complex protein from Helicobacter pylori's type IV secretion system (T4SS) and may participate in the modulation of gastric tissue inflammation. A three-dimensional structure has been reported for only two segments of the protein. To build a more complete model, particularly the region that spans between the outer membrane (OM) and the inner membrane (IM), we employed different approaches, including homology modeling, ab initio, and deep learning techniques. For the long-middle repeat region (MRR), modeling was performed using deep learning techniques and Molecular Dynamics. The modeled segments were assembled into a chain of 1595 aa, and a 14-chain CagY multimer structure was composed by structural alignment. The final multimer structure correlated with previously published struc-tures and allows to show how the multimer may form the T4SS channel through which CagA and other molecules are translocated to gastric epithelial cells. The model further confirmed that MRR, the most polymorphic and complex region of CagY, presents numerous cysteine residues forming disulfide bonds that stabilize the protein and suggest this domain probably functions as a contractile region that may play an essential role in the modulatory activity of CagY on tissue inflammation.

Conservation tillage and deep side-fertilization both hold the potential to reduce nitrogen leaching and improve grain yield and nitrogen use efficiency in fragrant rice cultivation practices. However, the combined impact of different tillage practices with deep side-fertilization on nitrogen leaching remains uncertain. Therefore, this study conducted on-site experiments for four rice growing seasons in both early and late seasons in 2018 and 2019, using the fragrant rice varieties "Meixiangzhan 2"(MX) and "Xiangyaxiangzhan"(XY). The four experimental treatments included:conventional tillage with regular fertilization (T1), conventional tillage with simultaneous deep fertilization (T2), reduced tillage with simultaneous deep fertilization (T3) , and no-tillage with simultaneous deep fertilization (T4). Our results indicate that the T4 treatment exhibited higher nitrogen leaching rates and potential nitrogen losses throughout the entire rice growth cycle, with a 4.51% increase in total nitrogen leaching and a 1.86% increase in potential nitrogen leaching compared to T1. In contrast, the T2 treatment demonstrated the lowest nitrogen leaching rate, resulting in a 6.01% reduction in total nitrogen leaching and a 9.57% decrease in potential nitrogen leaching compared to T1, demonstrating the most optimal performance. For MX, the T1 treatment resulted in lower daily grain outputs compared to the other treatments, with disparities ranging from 5.35% to 9.94%. Similarly, for the XY, the T1 treatment yielded significantly lower daily grain outputs compared to the other treatments, with discrepancies ranging from 6.26% to 10.81% during the late season of 2019. Overall, this study shows that conventional tillage with deep fertilization effectively reduces nitrogen leaching and boosts fragrant rice yields. Minimizing nitrogen leaching enhances nitrogen utilization efficiency. These insights highlight the significance of strategic fertilization and cultivation for environmental conservation and agricultural sustainability.

All kinds of vessels consist of dozens of complex machineries with rotating parts and electric motors that operate continuously in a harsh environment with excess temperature and humidity, vibration, fatigue and load. A breakdown or malfunction in one of these machineries can significantly impact the vessel’s operation and safety and consequently, the safety of the crew and the environment. To maintain operational efficiency and seaworthiness, the shipping industry invests substantial resources in preventive maintenance and repairs. This research presents the economic and technical benefits of predictive maintenance over to traditional preventive maintenance, and repair by replacement approaches in the maritime domain. By leveraging modern technology and Artificial Intelligence, we can analyze real-time operating conditions of machinery, enabling early detection of potential damages and allowing for effective planning of future maintenance and repair activities. In this paper, we propose and develop a Convolutional Neural Network that is fed with raw vibration measurements acquired in a laboratory environment from the ball bearings of a motor. Then, we investigate whether the proposed network can accurately detect the functional state of ball bearings and categorize any possible failures present, contributing to improved maintenance practices in the shipping industry.

Digital Microfluidic Biochips (DMFBs), used in various kinds of fields like DNA analysis, clinical diagnosis, and PCR testing, have made biochemical experiments more compact, efficient, and user-friendly than previous ways. However, their reliability is often compromised by their inability to adapt to all kinds of errors. All errors in biochips can be categorized into two types: known errors and unknown errors. Known errors are detectable before the start of the routing process through sensors or cameras. Unknown errors, in contrast, become apparent only during the routing process and remain undetected by sensors or cameras, which is the biggest issue to unexpectedly stop the routing process and diminishes the reliability of biochips. This paper introduces a deep reinforcement learning-based routing algorithm designed to manage not only known errors but also unknown errors. Our experiments demonstrate that our algorithm outperforms previous ones in terms of the success rate of the routing in the scenario including both known errors and unknown errors. Additionally, our algorithm contributes to detecting unknown errors during the routing process and identifying the most efficient routing path with high probability.

This study proposes a method for image matching between infrared(IR)-RGB images using a deep learning network to estimate the deformation field. We propose a deformation field generator (DFG) that estimates the deformation field of the transformation matrix to match each pixel or IR image to the RGB image. DFG is a network that receives IR and RGB images as input; the output size is two channels and has the sample resolution as the input image. By warping the IR image through a grid-sampler that warps the image according to the value of the deformation field, we can obtain a warped IR image that matches the RGB image. Additionally, to check whether the warped IR image matched the RGB image, the masking images detecting the segmentation of objects were photographed in two images. Without directly comparing IR and RGB images, we proposed mask loss that warps the IR mask image through the deformation field and grid sampler and then compares the warped IR mask image with the RGB mask image. Mask loss solves the spatial similarity comparison problem with multi-modality images, such as IR and RGB images, by comparing the mask image with the same modality image as the mask image.

Vehicle re-identification research under surveillance cameras has yielded impressive results. However, the challenge of Unmanned Aerial Vehicle (UAV)-based vehicle re-identification (ReID) presents a high degree of flexibility, mainly due to complicated shooting angles, occlusions, low discrimination of top-down features, and significant changes in vehicle scales. To address this, we propose a novel Dual Mixing Attention Network (DMANet) to extract discriminative features robust to variations in viewpoint. Specifically, we first present a plug-and-play Dual Mixing Attention Module (DMAM) to capture pixel-level pairwise relationships and channel dependencies, where DMAM is composed of Spatial Mixing Attention (SMA) and Channel Mixing Attention (CMA). First, the original feature is divided according to the dimensions of spatial and channel to obtain multiple subspaces. Then, a learnable weight is applied to capture the dependencies between local features in the mixture space. Finally, the features extracted from all subspaces are aggregated to promote their comprehensive feature interaction. Moreover, the DMAM can be readily inserted into backbone networks at any depth to improve vehicle discrimination. The experiments show that the proposed structure gains a better performance compared with the representative methods in the UAV-based vehicle ReID task. Our code and models will be publicly released.

Chlorophyll-a (Chla) is a crucial pigment in phytoplankton, playing a vital role in determining phytoplankton biomass and water nutrient status. However, in optically complex water bodies, Chla concentration is no longer the primary factor influencing remote sensing spectral reflectance signals, leading to significant errors in traditional Chla concentration estimation methods. With advancements in in-situ measurements, synchronized satellite data, and computer technology, machine learning algorithms have become popular in Chla concentration retrieval. Nevertheless, when using machine learning methods to estimate Chla concentration, abrupt changes in Chla values can disrupt the spatiotemporal smoothness of the retrieval results. Therefore, this study proposes a two-stage approach to enhance the accuracy of Chla concentration estimation in optically complex water bodies. In the first stage, a one-dimensional convolutional neural network (1DCNN) is employed for precise Chla retrieval, and in the second stage, the regression layer of the 1DCNN is replaced with Support Vector Regression (SVR). The research findings are as follows: (1) In the first stage, the performance metrics (R², RMSE, RMLSE, Bias, MAE) of the 1DCNN outperform state-of-the-art algorithms (OCI, SVR, RFR) on the test dataset. (2) After the second stage, the performance further improves, with the metrics achieving values of 0.892, 11.243, 0.052, 1.056, and 1.444, respectively. (3) In mid-to-high latitude regions, the inversion performance of 1DCNN\SVR is superior to other algorithms, exhibiting richer details and higher noise tolerance in nearshore areas. (4) 1DCNN\SVR demonstrates high inversion capabilities in water bodies with medium to high nutrient levels.

Deep reinforcement learning is one of the research hotspots in artificial intelligence and has been successfully applied in many research areas, however, the low training efficiency and high demand for samples are problems that limit the application To address these problems, a hierarchical episodic control model extending episodic memory to the domain of hierarchical reinforcement learning is proposed in this paper. The model is theoretically justified and employs a hierarchical implicit memory planning approach for counterfactual trajectory value estimation. Starting from the final step and recursively moving back along the trajectory, a hidden plan is formed within the episodic memory. Experience is aggregated both along trajectories and across trajectories, and the model is updated using a multi-headed backpropagation similar to bootstrapped neural networks. This model extends the parameterized episodic memory framework to the realm of hierarchical reinforcement learning and is theoretically analyzed to demonstrate its convergence and effectiveness. Experiments conducted in Four Room, Mujoco, and UE4-based active tracking , highlight that the hierarchical episodic control model effectively enhances training efficiency. It demonstrates notable improvements in both low-dimensional and high-dimensional environments, even in cases of sparse rewards.

Retailers must have accurate sales forecasts to operate their businesses efficiently and effectively and to remain competitive in the marketplace. Global forecasting models like RNNs can be a powerful tool for forecasting in retail settings, where multiple time series are often interrelated and influenced by a variety of external factors. By including covariates in a forecasting model, we can better capture the various factors that can influence sales in a retail setting. This can help improve the accuracy of our forecasts and enable better decision-making for inventory management, purchasing, and other operational decisions. In this study we investigate how the accuracy of global forecasting models is affected by the inclusion of different potential demand covariates. To ensure the significance of the study's findings, we used the M5 forecasting competition's openly accessible and well-established dataset. The results obtained from the DeepAR models, which were trained on various combinations of features including time, price, events, and IDs, suggest that individually only the features corresponding to IDs improve the baseline model. However, when all the features are used together, the best performance is achieved, indicating that the individual relevance of each feature is emphasized when the information is given jointly. Comparing the model with features to the model without features, there is an improvement of 1.76\% for MRMSSE and 6.47\% for MMASE.

This paper explores the potential of using the SAM segmentator to enhance the segmentation capability of known methods. SAM is a promptable segmentation system that offers zero-shot generalization to unfamiliar objects and images, eliminating the need for additional training. The open-source nature of SAM on GitHub allows for easy access and implementation. In our experiments, we aim to improve the segmentation performance by providing SAM with checkpoints extracted from the masks produced by DeepLabv3+, then merging the segmentation masks provided by these two networks. Additionally, we examine the \enquote{oracle} method (as upper bound baseline performance), where segmentation masks are inferred only by SAM with checkpoints extracted from ground truth. In addition, we tested in the CAMO datasets an ensemble of PVTv2 transformers; combining the ensemble and SAM yields state-of-the-art performance in that dataset. The results of our study provide valuable insights into the potential of incorporating the SAM segmentator into existing segmentation techniques. We release with this paper the open-source implementation of our method.

Fractures of the talar neck and body are associated with spine fractures and scoliosis deformity, which affect cosmetic appearance and cause difficulty in ambulation. The implant design for talus surgery is thriving as a functional alternative in case of severe talar destruction, focusing on segmentation and reconstruction of the talus’s shape. However, manual segmentation of the talus is time-consuming and subjective. In this study we exploited the automatic segmentation framework to efficiently train a deep learning-based model to accurately segment the talus based on computed tomography imaging. We developed three model configurations with nnU-Net and investigated their Dice similarity coefficients (DSC) and 95% Hausdorff distances (HD95) for talus segmentation on a CT image dataset. The three configurations performed well (DSC > 0.95, HD95 < 0.6). When tested on the same samples, one of the configurations was more efficient while ensuring higher accuracy. We propose to focus on talus anatomic variations with increasing age based on this framework and apply it to clinical trials at the next stage.

Buildings are responsible for approximately 40% of the world’s energy consumption and 36% of the total carbon dioxide emissions. Building occupancy is essential, enabling Occupant-Centric Control for zero emissions and decarbonization. Although existing machine learning and deep learning methods for building occupancy prediction have achieved remarked progress, their analyses remain limited when applied to complex real-world scenarios. Besides, there is a high expectation for Transformer algorithms to predict building occupancy accurately. Therefore, this paper presents an Occupancy Prediction Transformer network (OPTnet). We fuse and feed multi-sensor data (building occupancy, indoor environmental conditions, HVAC operations) into a Transformer model to forecast the future occupancy presence in multiple zones. We perform experimental analysis and compare it to different occupancy prediction methods (e.g., Decision Tree, Long Short-Term Memory networks, Multi-layer perceptron) and diverse time horizons (1,2,3,5,10,20,30mins). The performance metrics (e.g., Accuracy and Mean Squared Error) are employed to evaluate the effectiveness of prediction algorithms. Our OPTnet method achieves superior performance on our experimental two-week data compared to existing methods. The improved performance signifies its potential to enhance HVAC control systems and energy optimization strategies. We will make the code publicly available at https://github.com/kailaisun/occupancy-prediction-binary to promote transparency and reproducibility.

GNSS time series prediction plays a significant role in monitoring crustal plate motion, landslide detection, and maintenance of the global coordinate framework. Long Short-Term Memory (LSTM), a deep learning model has been widely applied in the field of high-precision time series prediction especially when combined with Variational Mode Decomposition (VMD) to form the VMD-LSTM hybrid model. To further improve the prediction accuracy of the VMD-LSTM model, this paper proposes a dual variational modal decomposition long short-term memory (DVMD-LSTM) model to effectively handle the noise in GNSS time series prediction. This model extracts fluctuation features from the residual terms obtained after VMD decomposition to reduce the prediction errors associated with residual terms in the VMD-LSTM model. Daily E, N, and U coordinate data recorded at multiple GNSS stations between 2000 and 2022 are used to validate the performance of the proposed DVMD-LSTM model. The experimental results demonstrate that compared to the VMD-LSTM model, the DVMD-LSTM model achieves significant improvements in prediction performance across all measurement stations. The average RMSE is reduced by 9.86%, and the average MAE is reduced by 9.44%. Furthermore, the average accuracy of the optimal noise model for the predicted results is improved by 36.50%, and the average speed accuracy of the predicted results is enhanced by 33.02%. These findings collectively attest to the superior predictive capabilities of the DVMD-LSTM model, thereby enhancing the reliability of the predicted results.

Time series (TS)-based predictions are made by examining the behaviour of historical data to forecast future values. Multiple industries; such as stock market trading, power load forecasting, medical monitoring, and intrusion detection; frequently use it. Several variables; such as the performance of other markets as well as the economic situation of a country which affect its market performance; significantly affect the prediction of stock market prices. Therefore, this present study uses numerous variables; such as the opening, lowest, highest, and closing prices; to predict the indices of the stock market of the Kingdom of Saudi Arabia (KSA). Successfully accomplishing an investment goal largely depends on choosing the right stocks to buy, sell, or maintain. The project's output is the projected closing prices (regression) over the next seven days, which helps investors make the best decisions. This present study used exponential smoothing (ES) to remove noise from the input data obtained from the Saudi Stock Exchange, or Tadawul, before using a multivariate long short-term memory (LSTM) deep learning (DL) algorithm to forecast stock market prices. The proposed multivariate LSTMDL model had satisfactory prediction rates of 97.49% and 92.19% for the univariate model. Therefore, it can be used to effectively predict stock market prices. The results also indicate that DL as well as multiple sources of information can be used to accurately predict stock markets.

Sentiment analysis of public opinion expressed in social networks has been developed into various applications, especially in English. Hybrid approaches are potential models for reducing sentiment errors on increasingly complex training data. This paper aims to test some hybrid deep learning models' reliability in some domains' Vietnamese language. Our research questions are to determine whether it is possible to produce hybrid models that outperform the Vietnamese language. Hybrid deep sentiment-analysis learning models are built and tested on reviews and feedback of the Vietnamese language. The hybrid models outperformed the accuracy of Vietnamese sentiment analysis on Vietnamese datasets. It contributes to the growing body of research on Vietnamese NLP, providing insights and directions for future studies in this area.

The ability to handle spatiotemporal information makes contribution for improving the prediction performance of machine RUL. However, most existing models for spatiotemporal information processing are not only complex in structure but also lack adaptive feature extraction capabilities. Therefore, a lightweight operator with adaptive spatiotemporal information extraction ability named Involution GRU (Inv-GRU) is proposed for aero-engine RUL prediction. Involution, the adaptive feature extraction operator, is replaced by the information connection in the gated recurrent unit for obtaining the adaptively spatiotemporal information extraction ability and reducing the parameters. Thus, Inv-GRU can well extract the degradation information the of aero-engine. Then for RUL prediction task, the Inv-GRU based deep learning (DL) framework is firstly constructed, where features extracted by Inv-GRU and several human-made features are separately processed to generate the health indicators (HIs) from multi-raw data of aero-engines. Finally, fully connection layers are adopted are adopted to reduce dimension and regress RUL based on the generated HIs. By applying the Inv-GRU based DL framework to the Commercial Modular Aero Propulsion System Simulation (C-MAPSS) datasets, successful predictions of aero-engines RUL have been achieved. Comparative analysis reveals that the proposed model exhibits superior overall prediction performance compared to recent public methods.

The COVID-19 pandemic has affected millions of people globally, with respiratory organs being strongly affected in individuals with comorbidities. Medical imaging-based diagnosis and prognosis have become increasingly popular in clinical settings to detect COVID-19 lung infections. Among various medical imaging modalities, ultrasound stands out as low-cost, mobile, and radiation-safe imaging technology. In this comprehensive review, we focus on ultrasound-based AI studies for COVID-19 detection that use public or private lung ultrasound datasets. We surveyed articles that used publicly available lung ultrasound datasets for COVID-19 and reviewed publicly available datasets and organize ultrasound-based AI studies per dataset. We analyzed and tabulated studies in several dimensions, such as data preprocessing, AI models, cross-validation, and evaluation criteria. In total, we reviewed 42 articles, where 28 articles used public datasets, and the rest used private data. Our findings suggest that ultrasound-based AI studies for the detection of COVID-19 have great potential for clinical use, especially for children and pregnant women. Our review also provides a useful summary for future researchers and clinicians who may be interested in the field.

Selecting the optimal activation function for training deep neural networks has always been challenging because it significantly impacts the neural network’s performance and training speed. At this point, researchers are more likely to employ RELU than other well-known activation functions. After RELU, researchers have proposed many activation functions to improve RELU. None of them was capable of surpassing RELU as their most significant rival. SWISH outperformed RELU in several challenging experiments like classification, object detection, and tracking. Replacing RELU units with SWISH, which improves performance in many tasks. This paper proposed a new activation function surpassing Google’s brain’s SWISH function, Which we named AIF. Experiments indicate that our proposed activation function outperforms SWISH in various tasks.

In urban areas, deep excavation-induced ground deformation may damage adjacent existing structures and is conventionally evaluated by levelling at installed settlement points. However, a small number of measurements cannot represent the total change in ground deformation adjacent to excavation sites. Furthermore, significant local subsidence may occur in places where settlement points have not been installed and only noticed after an accident. For deep excavation sites located in urban areas, paved pedestrian sidewalks are often located adjacent to sites, and construction activity can cause these paving blocks to displace. This study introduces a method to detect paving block displacement adjacent to deep excavation sites using terrestrial photogrammetry. A digital camera creating point cloud data (PCD) and an acquisition method satisfying the frontal and side overlap requirements were demonstrated. To investigate the displacement detection and measurement capabilities by PCD analysis, an experimental program was conducted, including a PCD comparison containing the uplift, settlement, and horizontal paving block displacement and reference data. The cloud-to-cloud distance computation algorithm was adopted for PCD comparison. Paving block displacement was detected for displacements of 5, 7.5, and 10 mm in the uplift, settlement, and horizontal directions; however, the horizontal displacements were less clear. PCD analysis enabled satisfactory measurements between 0.024 and 0.881 mm for the vertical-displacement cases, but a significant error was observed for the horizontal-displacement cases owing to the cloud-comparison algorithm. The measurement blind spot of limited settlement points was overcome by the proposed method that detected and measured paving block displacement adjacent to excavation sites.

Computer vision methods have shown to be effective in classifying garbage into recycling categories for waste processing, existing methods are costly, imprecise, and unclear. To tackle this issue, we introduce MWaste, a mobile application that uses computer vision and deep learning techniques to classify waste materials as trash, plastic, paper, metal, glass or cardboard. Its effectiveness was tested on various neural network architectures and real world images, achieving an average precision of 92% on the test set. This app can help combat climate change by enabling efficient waste processing and reducing the generation of greenhouse gases caused by incorrect waste disposal.

Auto-labeling is one of the main challenges in 3D vehicle detection. Auto-labeled datasets can be used to identify objects in LiDAR data, which is a challenging task due to the large size of the dataset. In this work, we propose a novel methodology to generate new 3D based auto-labeling datasets with a different point of view setup than the one used in the most recognized datasets (KITTI, WAYMO, etc.). The performance of the methodology has been further demonstrated with the development of our own dataset with the auto-generated labels and tested under boundary conditions on a bridge in a fixed position. The proposed methodology is based on the YOLO model trained with the KITTI dataset. From a camera-LiDAR sensory fusion, it is intended to auto-label new datasets while maintaining the consistency of the Ground Truth. The main contribution of this work is a novel methodology to auto-label autonomous driving datasets using YOLO as the main labelling system. The performance of this approach is measured retraining the contrast models of the KITTI benchmark.

Sometimes when diabetic retinopathy (DR) is found and treated quickly, vision loss can indeed be spared. This study deploys a deep learning (DL) model that can discover all 5 stages of DR more accurately than other methods. The proposed methodology shows two cases scenarios: case 1 with image enhancement using CLAHE and ESRGAN, and case 2 without image enhancement. Augmentation techniques are then employed to produce a balanced dataset with the identical criteria for both scenarios. The generated model using DenseNet-121 on the APTOS dataset outperformed other approaches for locating the 5 stages of DR, with an accuracy of 98.7 percent for case 1 and 81.2 percent for case 2. Using CLAHE and ESRGAN was shown to improve a model's performance and ability to learn.

Deep learning technologies for skin cancer detection have been dramatically advanced based on high resolution dermoscopic images. The low-cost approach based on clinical skin images of low resolution is promising but remains technically challenging due to their undermined image quality. In this paper, we propose a coarse-to-fine efficient super resolution transformer (CF-ESRT) network to reconstruct the dermoscopy-level high resolution skin image from a low resolution clinical image. By connecting the refinement network to the original super resolution transformers and applying perceptual and gradient losses, our framework noticeably improves the finer texture details of skin lesions in the super resolution (SR) images, and is effective to elevate the perceptual quality of the SR images. Quantitative and qualitative evaluations show that our method outperforms ESRT the basis model as well as the other state-of-the-art SR models.

Pulsed thermography is a vital technique in the nondestructive evaluation field. However, its data analysis can be complex and requires skilled experts. Advances in deep learning have yielded exceptional results, including image segmentation. Therefore, many efforts have been made to apply deep learning methods to data processing for nondestructive evaluation. Despite this, there is currently no public Pulsed thermographic dataset available for evaluating various spatial-temporal deep methods of segmenting pulsed thermographic data. This article aims to provide such a dataset and assess the performance of commonly used deep learning-based instance segmentation models on it. Additionally, the impact of the number of frames and data transformations on model performance is examined. The findings suggest that suitable preprocessing methods can effectively reduce the data size without compromising the deep models’ performance.