Application Of Deep Learning Models Research Articles

Identifying autism spectrum disorder from multi-modal data with privacy-preserving

The application of deep learning models to precision medical diagnosis often requires the aggregation of large amounts of medical data to effectively train high-quality models. However, data privacy protection mechanisms make it difficult to perform medical data collection from different medical institutions. In autism spectrum disorder (ASD) diagnosis, automatic diagnosis using multimodal information from heterogeneous data has not yet achieved satisfactory performance. To address the privacy preservation issue as well as to improve ASD diagnosis, we propose a deep learning framework using multimodal feature fusion and hypergraph neural networks for disease prediction in federated learning (FedHNN). By introducing the federated learning strategy, each local model is trained and computed independently in a distributed manner without data sharing, allowing rapid scaling of medical datasets to achieve robust and scalable deep learning predictive models. To further improve the performance with privacy preservation, we improve the hypergraph model for multimodal fusion to make it suitable for autism spectrum disorder (ASD) diagnosis tasks by capturing the complementarity and correlation between modalities through a hypergraph fusion strategy. The results demonstrate that our proposed federated learning-based prediction model is superior to all local models and outperforms other deep learning models. Overall, our proposed FedHNN has good results in the work of using multi-site data to improve the performance of ASD identification.

Real-Time Violence Detection in Surveillance Videos Using Deep Learning Approach

Abstract: In order to tackle the ongoing problem of keeping fruits fresh, this research presents an automated freshness detection. The ever-growing presence of surveillance cameras has undoubtedly enhanced public safety. However, manually reviewing vast quantities of video footage for violence detection remains a tedious and error-prone task. This paper proposes a deep learningbased approach for anomaly detection in surveillance videos, with a specific focus on identifying violent activities. Traditional surveillance methods often lack the sophistication to distinguish between normal behaviour and potentially threatening actions. Deep learning offers a significant advantage by automating the identification of deviations from established behavioural patterns in video data. This automation enables real-time analysis of footage, potentially signifying the occurrence of violent incidents and allowing for a swifter response from security personnel. This research delves into the application of deep learning models for violence detection in surveillance footage. We begin by discussing the limitations inherent in conventional methods, such as motion detection and manual review, which struggle to capture the nuances of human behaviour. Subsequently, we explore the advantages that deep learning-based approaches bring to the table, including their ability to learn complex patterns from large datasets and identify subtle anomalies that might escape human observation.

Comparison of deep learning models for milk production forecasting at national scale

Forecasting of agricultural variables at national scale is essential for developing policies aimed at guaranteeing food security and increase the stability of the entire agri-food chain. However, this is often a challenging task due to the fact that data are usually sparse and long records are rarely available. This work aims at evaluating the potentiality of ten deep learning models for predicting monthly milk production in France, Germany, and Italy, using as input, climatic and economic variables from open datasets. The results indicate that deep learning models are a better alternative to traditional statistical approaches, providing robust results without requiring complex model architectures. Furthermore, the prominent autoregressive nature of milk production highlights the inability of environmental variables to capture external processes that influence the milk production (e.g., milk quota). However, the high accuracy achieved by the models lays the ground for possible application of deep learning models to accurately forecast agricultural variables at national scale with potential implications for the development of dairy insurance products, and risk management practices.

Multimodal Machine Learning in Image-Based and Clinical Biomedicine: Survey and Prospects

AbstractMachine learning (ML) applications in medical artificial intelligence (AI) systems have shifted from traditional and statistical methods to increasing application of deep learning models. This survey navigates the current landscape of multimodal ML, focusing on its profound impact on medical image analysis and clinical decision support systems. Emphasizing challenges and innovations in addressing multimodal representation, fusion, translation, alignment, and co-learning, the paper explores the transformative potential of multimodal models for clinical predictions. It also highlights the need for principled assessments and practical implementation of such models, bringing attention to the dynamics between decision support systems and healthcare providers and personnel. Despite advancements, challenges such as data biases and the scarcity of “big data” in many biomedical domains persist. We conclude with a discussion on principled innovation and collaborative efforts to further the mission of seamless integration of multimodal ML models into biomedical practice.

Deep learning prediction of renal anomalies for prenatal ultrasound diagnosis

Deep learning algorithms have demonstrated remarkable potential in clinical diagnostics, particularly in the field of medical imaging. In this study, we investigated the application of deep learning models in early detection of fetal kidney anomalies. To provide an enhanced interpretation of those models’ predictions, we proposed an adapted two-class representation and developed a multi-class model interpretation approach for problems with more than two labels and variable hierarchical grouping of labels. Additionally, we employed the explainable AI (XAI) visualization tools Grad-CAM and HiResCAM, to gain insights into model predictions and identify reasons for misclassifications. The study dataset consisted of 969 ultrasound images from unique patients; 646 control images and 323 cases of kidney anomalies, including 259 cases of unilateral urinary tract dilation and 64 cases of unilateral multicystic dysplastic kidney. The best performing model achieved a cross-validated area under the ROC curve of 91.28% ± 0.52%, with an overall accuracy of 84.03% ± 0.76%, sensitivity of 77.39% ± 1.99%, and specificity of 87.35% ± 1.28%. Our findings emphasize the potential of deep learning models in predicting kidney anomalies from limited prenatal ultrasound imagery. The proposed adaptations in model representation and interpretation represent a novel solution to multi-class prediction problems.

Evaluation of Approaches for Early Stroke Detection and Diagnosis Using EMG Data: Features, Techniques, and Challenges

This paper provides a thorough analysis of the use of electromyography (EMG) data in early stroke diagnosis and detection. Stroke continues to be a major global cause of disability and death, which emphasises the critical need for an accurate diagnosis made quickly to improve patient outcomes. Early detection is still difficult to achieve, even with improvements in medical imaging and testing technologies. By detecting minute variations in muscle activity linked to stroke symptoms, EMG data analysis offers a viable method for early stroke identification. The review delves into the diverse methodologies and strategies utilised to leverage EMG data for the purpose of stroke diagnosis, encompassing the application of deep learning models and machine learning algorithms. The paper proposes a structured framework for classifying approaches for early stroke detection and diagnosis using EMG data, providing a systematic way to categorize and compare different methodologies. The paper concludes by highlighting the revolutionary potential of EMG-based techniques in improving the diagnosis of strokes earlier and urging more study to address current issues and make clinical application easier.

Application of deep learning model incorporating domain knowledge in international migration forecasting

PurposeForecasting population movement trends is crucial for implementing effective policies to regulate labor force growth and understand demographic changes. Combining migration theory expertise and neural network technology can bring a fresh perspective to international migration forecasting research.Design/methodology/approachThis study proposes a conditional generative adversarial neural network model incorporating the migration knowledge – conditional generative adversarial network (MK-CGAN). By using the migration knowledge to design the parameters, MK-CGAN can effectively address the limited data problem, thereby enhancing the accuracy of migration forecasts.FindingsThe model was tested by forecasting migration flows between different countries and had good generalizability and validity. The results are robust as the proposed solutions can achieve lesser mean absolute error, mean squared error, root mean square error, mean absolute percentage error and R2 values, reaching 0.9855 compared to long short-term memory (LSTM), gated recurrent unit, generative adversarial network (GAN) and the traditional gravity model.Originality/valueThis study is significant because it demonstrates a highly effective technique for predicting international migration using conditional GANs. By incorporating migration knowledge into our models, we can achieve prediction accuracy, gaining valuable insights into the differences between various model characteristics. We used SHapley Additive exPlanations to enhance our understanding of these differences and provide clear and concise explanations for our model predictions. The results demonstrated the theoretical significance and practical value of the MK-CGAN model in predicting international migration.

One-shot skill assessment in high-stakes domains with limited data via meta learning

Deep Learning (DL) has achieved robust competency assessment in various high-stakes fields. However, the applicability of DL models is often hampered by their substantial data requirements and confinement to specific training domains. This prevents them from transitioning to new tasks where data is scarce. Therefore, domain adaptation emerges as a critical element for the practical implementation of DL in real-world scenarios. Herein, we introduce A-VBANet, a novel meta-learning model capable of delivering domain-agnostic skill assessment via one-shot learning. Our methodology has been tested by assessing surgical skills on five laparoscopic and robotic simulators and real-life laparoscopic cholecystectomy. Our model successfully adapted with accuracies up to 99.5 % in one-shot and 99.9 % in few-shot settings for simulated tasks and 89.7 % for laparoscopic cholecystectomy. This study marks the first instance of a domain-agnostic methodology for skill assessment in critical fields setting a precedent for the broad application of DL across diverse real-life domains with limited data.

Research and Application of Deep Learning in Medical Image Reconstruction and Enhancement

In recent years, deep learning technology has made remarkable progress in medical image reconstruction and enhancement, and has become one of the research hotspots in the field of medical image processing. This paper discusses the latest research progress and application of deep learning in medical image reconstruction and enhancement. Firstly, the importance of medical image reconstruction and enhancement and the limitations of traditional methods are introduced. Then, a detailed discussion was conducted on the application of deep learning models, including Convolutional Neural Networks (CNN), Generative Adversarial Networks (GAN), and Autoencoders, in medical image processing. Specifically, an analysis and comparison were conducted on the image reconstruction ability of CNN models, the image enhancement effect of GAN models, and the image denoising and reconstruction of Autoencoder models. Then, the advantages and challenges of deep learning model in medical image processing are discussed, and the future development direction is discussed. Finally, the research results of this paper are summarized and the prospect of future research is put forward. The research in this paper provides some enlightenment and reference for researchers and practitioners in the field of medical image processing, which is helpful to promote the continuous innovation and progress of medical image processing technology.

Deep learning-based retinal abnormality detection from OCT images with limited data

In the realm of medical diagnosis, the challenge posed by retinal diseases is considerable, given their potential to complicate vision and overall ocular health. A promising avenue for achieving highly accurate classifiers in detecting retinal diseases involves the application of deep learning models. However, overfitting issues often undermine the performance of these models due to the scarcity of image samples in retinal disease datasets. To address this challenge, a novel deep triplet network is proposed as a metric learning approach for detecting retinal diseases using Optical Coherence Tomography (OCT) images. Incorporating a conditional loss function tailored to the constraints of limited data samples, this deep triplet network enhances the model’s accuracy. Drawing inspiration from pre-trained models such as VGG16, the foundational architecture of our model is established. Experiments use open-access datasets comprising retinal OCT images to validate our proposed approach. The performance of the suggested model is demonstrated to surpass that of state-of-the-art models in terms of accuracy. This substantiates the effectiveness of the deep triplet network in addressing overfitting issues associated with limited data samples in retinal disease datasets.

Collaborative Application of Deep Learning Models for Enhanced Accuracy and Prediction in Carbon Neutrality Anomaly Detection

In the face of intensifying global climate change, carbon neutrality has emerged as a pivotal strategy to curb greenhouse gas emissions and confront the complexities associated with climate challenges. However, achieving carbon neutrality poses a formidable challenge: the identification and mitigation of anomalies within the carbon sequestration process. These anomalies can result in unintended carbon dioxide leakage, emissions, or system failures, thus jeopardizing the feasibility and resilience of carbon neutrality initiatives. This research introduces the ResNet-BIGRU-TPA network, an innovative model that integrates deep learning techniques with time series attention mechanisms. The primary focus centers on addressing the intricate task of anomaly detection within the realm of carbon offsetting, specifically aiming to enhance precision in identifying a wide array of complex anomalous events. Through rigorous experimental validation across four diverse datasets, the model has exhibited exceptional performance.

THAR- Targeted Hate Speech Against Religion: A high-quality Hindi-English code-mixed Dataset with the Application of Deep Learning Models for Automatic Detection

During the last decade, social media has gained significant popularity as a medium for individuals to express their views on various topics. However, some individuals also exploit the social media platforms to spread hatred through their comments and posts, some of which target individuals, communities or religions. Given the deep emotional connections people have to their religious beliefs, this form of hate speech can be divisive and harmful, and may result in issues of mental health as social disorder. Therefore, there is a need of algorithmic approaches for the automatic detection of instances of hate speech. Most of the existing studies in this area focus on social media content in English, and as a result several low-resource languages lack computational resources for the task. This study attempts to address this research gap by providing a high-quality annotated dataset designed specifically for identifying hate speech against religions in the Hindi-English code-mixed language. This dataset “Targeted Hate Speech Against Religion” (THAR)) consists of 11,549 comments and has been annotated by five independent annotators. It comprises two subtasks: (i) Subtask-1 (Binary classification), (ii) Subtask-2 (multi-class classification). To ensure the quality of annotation, the Fleiss Kappa measure has been employed. The suitability of the dataset is then further explored by applying different standard deep learning, and transformer-based models. The transformer-based model, namely Multilingual Representations for Indian Languages (MuRIL), is found to outperform the other implemented models in both subtasks, achieving macro average and weighted average F1 scores of 0.78 and 0.78 for Subtask-1, and 0.65 and 0.72 for Subtask-2, respectively. The experimental results obtained not only confirm the suitability of the dataset but also advance the research towards automatic detection of hate speech, particularly in the low-resource Hindi-English code-mixed language.

Artificial Intelligence in Intellectual Property Protection: Application of Deep Learning Model

To create and train a deep learning model costs a lot in comparison to ascertain a trained model. So, a trained model is considered as the intellectual property (IP) of the person who creates such model. However, there is every chance of illegal copying, redistributing and abusing of any of these high-performance models by the malicious users. To protect against such menaces, a few numbers of deep neural networks (DNN) IP security techniques have been developed recently. The present study aims at examining the existing DNN IP security activities. In the first instance, there is a proposal of taxonomy in favor of DNN IP protection techniques from the perspective of six aspects such as scenario, method, size, category, function, and target models. Afterwards, this paper focuses on the challenges faced by these methods and their capability of resisting the malicious attacks at different levels by providing proactive protection. An analysis is also made regarding the potential threats to DNN IP security techniques from various perspectives like modification of models, evasion and active attacks.
 Apart from that this paper look into the methodical assessment. The study explores the future research possibilities on DNN IP security by considering different challenges it would confront in the process of its operations.
 Result Statement: A high-performance deep neural Networks (DNN) model is costlier than the trained DNN model. It is considered as an intellectual property (IP) of the person who is responsible for creating DNN model. The infringement of the Intellectual Property of DNN model is a grave concern in recent years. This article summarizes current DNN IP security works by focusing on the limitations/ challenges they confront. It also considers the model in question's capacity for protection and resistance against various stages of attacks.

INTEGRATION OF DEEP LEARNING MODELS FOR VEHICLE COUNTING: TOWARDS OPTIMIZED PLANNING OF URBAN CHARGING INFRASTRUCTURES

Abstract. Amidst the rising prevalence of electric vehicles (EV) and the corresponding need for adequate charging infrastructures, this study presents an innovative application of deep learning models to analyze urban traffic. Utilizing the YOLO v3 model, our research focuses on developing a prototype for counting vehicles at an urban roundabout connecting three main roads, providing essential data for planning EV charging infrastructures. This work involves adapting and optimizing the YOLO v3 model to accurately identify and count vehicles passing through the roundabout. The aim is to measure traffic volume and identify circulation trends, which are crucial for understanding the spatial and temporal distribution of vehicles. The collected data not only estimate the number of EVs likely to traverse this area but also predict the needs for charging infrastructure in and around this specific roundabout. Our prototype demonstrated high accuracy in vehicle counting, allowing detailed analysis of traffic patterns at different times of the day. This analysis reveals significant variations in traffic flows, providing valuable insights into the most suitable times and locations for installing EV charging stations. While this study focuses on a single roundabout, it establishes a precedent for using advanced deep learning techniques in assessing EV charging infrastructure needs in broader urban contexts. The results suggest that such methods can be extended to other urban areas for more comprehensive and effective planning of EV charging infrastructures, thus contributing to more sustainable and intelligent urban mobility management.

Deep Learning Based Underwater Acoustic Target Recognition: Introduce a Recent Temporal 2D Modeling Method.

In recent years, the application of deep learning models for underwater target recognition has become a popular trend. Most of these are pure 1D models used for processing time-domain signals or pure 2D models used for processing time-frequency spectra. In this paper, a recent temporal 2D modeling method is introduced into the construction of ship radiation noise classification models, combining 1D and 2D. This method is based on the periodic characteristics of time-domain signals, shaping them into 2D signals and discovering long-term correlations between sampling points through 2D convolution to compensate for the limitations of 1D convolution. Integrating this method with the current state-of-the-art model structure and using samples from the Deepship database for network training and testing, it was found that this method could further improve the accuracy (0.9%) and reduce the parameter count (30%), providing a new option for model construction and optimization. Meanwhile, the effectiveness of training models using time-domain signals or time-frequency representations has been compared, finding that the model based on time-domain signals is more sensitive and has a smaller storage footprint (reduced to 30%), whereas the model based on time-frequency representation can achieve higher accuracy (1-2%).

Predicting Melting Temperatures of Freeze Valve in Molten Salt Reactors using LSTM-Based Deep Learning Models

The realm of molten salt reactors (MSR) presents unique challenges and opportunities in the pursuit of advancing nuclear energy technology. One pivotal aspect is the development of safety mechanisms, including freeze valves, aimed at enhancing the security and operational integrity of these reactors. Freeze valves, designed to halt the flow of molten salt in emergency scenarios, play a critical role in averting potential hazardous situations. This research paper investigates the application of deep learning models for predicting melting temperatures, leveraging time-series temperature data, specifically targeting the freeze valve technology within the context of molten salt reactors. The research methodology encompasses the creation of a synthetic dataset representative of freeze valve conditions in MSR. Each dataset entry comprises a time-series temperature profile associated with the freeze valve and its corresponding melting temperature. To capture the intricate temporal relationships inherent in these profiles, a specialized Recurrent Neural Network (RNN) architecture, incorporating Long Short-Term Memory (LSTM) cells, is developed and trained. The model’s training is augmented by preprocessing techniques, including the normalization of time-series temperature data. Evaluation of the model’s predictive capabilities focuses on its performance in forecasting the melting temperatures of the freeze valve. Metrics such as mean squared error (MSE) and mean absolute error (MAE) are utilized to quantify the model’s accuracy and robustness. The study delves into the influence of architectural configurations, hyperparameter tuning, and dataset characteristics on the model’s predictions, shedding light on the factors that contribute to optimal predictive performance. The outcomes of this research highlight the potential of deep learning models in predicting the melting temperatures of freeze valves in the context of MSR. The developed model not only showcases an ability to anticipate impending safety-critical events but also underscores the significance of leveraging time-series temperature data to enhance the integrity of safety mechanisms in nuclear reactor systems. Ultimately, this research contributes to the pursuit of safer and more efficient nuclear energy solutions, offering valuable insights into predictive modeling techniques within the complex landscape of molten salt reactor technologies.

Learning to Generalize over Subpartitions for Heterogeneity-Aware Domain Adaptive Nuclei Segmentation

Abstract Annotation scarcity and cross-modality/stain data distribution shifts are two major obstacles hindering the application of deep learning models for nuclei analysis, which holds a broad spectrum of potential applications in digital pathology. Recently, unsupervised domain adaptation (UDA) methods have been proposed to mitigate the distributional gap between different imaging modalities for unsupervised nuclei segmentation in histopathology images. However, existing UDA methods are built upon the assumption that data distributions within each domain should be uniform. Based on the over-simplified supposition, they propose to align the histopathology target domain with the source domain integrally, neglecting severe intra-domain discrepancy over subpartitions incurred by mixed cancer types and sampling organs. In this paper, for the first time, we propose to explicitly consider the heterogeneity within the histopathology domain and introduce open compound domain adaptation (OCDA) to resolve the crux. In specific, a two-stage disentanglement framework is proposed to acquire domain-invariant feature representations at both image and instance levels. The holistic design addresses the limitations of existing OCDA approaches which struggle to capture instance-wise variations. Two regularization strategies are specifically devised herein to leverage the rich subpartition-specific characteristics in histopathology images and facilitate subdomain decomposition. Moreover, we propose a dual-branch nucleus shape and structure preserving module to prevent nucleus over-generation and deformation in the synthesized images. Experimental results on both cross-modality and cross-stain scenarios over a broad range of diverse datasets demonstrate the superiority of our method compared with state-of-the-art UDA and OCDA methods. Graphical abstract

Predicting customer abandonment in recurrent neural networks using short-term memory

Customer retention, a critical business priority, has become a growing concern, especially in the telecommunications industry. This study addresses the need to anticipate and understand customer churn through the application of Deep Learning models. The central focus of the research was the development and evaluation of a short-term memory model (LSTM) specifically designed to predict customer leakage. The choice of LSTM as the mainstay of the research is based on its proven ability to model long-term dependencies in sequences, its resilience to recurrent challenges in neural networks, and its success in various sequence prediction tasks. The model implementation, configured sequentially with Keras, comprised of an initial LSTM layer of 64 units, followed by a 20% removal layer to mitigate overfitting. The second LSTM layer, with 32 units, was supplemented with another elimination layer. Model training was conducted using a dataset consisting of 20 attributes and 4250 records. The model evaluation was based on crucial measures such as precision, accuracy, sensitivity and F1 count, revealing exceptional results with 95% performance on all metrics. This study, therefore, highlights the effectiveness of the LSTM model in predicting customer churn, providing companies with a valuable tool to improve retention and mitigate associated losses.

Breast cancer detection using ensemble of convolutional neural networks

Early detection leading to timely treatment in the initial stages of cancer may decrease the breast cancer death rate. We propose deep learning techniques along with image processing for the detection of tumors. The availability of online datasets and advances in graphical processing units (GPU) have promoted the application of deep learning models for the detection of breast cancer. In this paper, deep learning models using convolutional neural network (CNN) have been built to automatically classify mammograms into benign and malignant. Issues like overfitting and dataset imbalance are overcome. Experimentation has been done on two publicly available datasets, namely mammographic image analysis society (MIAS) database and digital database for screening mammography (DDSM). Robustness of the models is accomplished by merging the datasets. In our experimentation, MatConvNet has achieved an accuracy of 94.2% on the merged dataset, performing the best amongst all the CNN models used individually. Hungarian optimization algorithm is employed for selection of individual CNN models to form an ensemble. Ensemble of CNN models led to an improved performance, resulting in an accuracy of 95.7%.

Meta-reservoir computing for learning a time series predictive model of wind power

Wind energy has become an essential part of the energy power source of current power systems since it is eco-friendly and sustainable. To optimize the operations of wind farms with the constraint of satisfying the power demand, it is critical to provide accurate predictions of wind power generated in the future. Although deep learning models have greatly improved prediction accuracy, the overfitting issue limits the application of deep learning models trained under one condition to another. A huge number of data are required to train a new deep learning model for another environment, which is sometimes not practical in some urgent situations with only very little training data on wind power. In this paper, we propose a novel learning method, named meta-reservoir computing (MRC), to address the above issue, combining reservoir computing and meta-learning. The reservoir computing method improves the computational efficiency of training a deep neural network for time series data. On the other hand, meta-learning is used to improve the initial point and other hyperparameters of reservoir computing. The proposed MRC method is validated using an experimental dataset of wind power compared with the traditional training method. The results show that the MRC method can obtain an accurate predictive model of wind power with only a few shots of data.