Deep Learning Processing Research Articles

An Empirical Study on Lightweight CNN Models for Efficient Classification of Used Electronic Parts

The problem of electronic waste (e-waste) presents a significant challenge in our society as outdated electronic devices are frequently discarded rather than recycled. To tackle this issue, it is important to embrace circular economy principles. One effective approach is to desolder and reuse electronic components, thereby reducing waste buildup. Automated vision-based techniques, often utilizing deep learning models, are commonly employed to identify and locate objects in sorting applications. Artificial intelligence (AI) and deep learning processes often require significant computational resources to perform automated tasks. These computational resources consume energy from the grid. Consequently, a rise in the use of AI can lead to higher demand for energy resources. This research empirically develops a lightweight convolutional neural network (CNN) model by exploring models utilising various grayscale image resolutions and comparing their performance with pre-trained RGB image classifier models. The study evaluates the lightweight CNN classifier’s ability to achieve an accuracy comparable to pre-trained red–green–blue (RGB) image classifiers. Experiments demonstrate that lightweight CNN models using 100 × 100 pixels and 224 × 224 pixels grayscale images can achieve accuracies on par with more complex pre-trained RGB classifiers. This permits the use of reduced computational resources for environmental sustainability.

A Review of AI-powered Diagnosis of Rare Diseases

The diagnosis of rare diseases presents significant challenges due to their low prevalence, complex symptomatology, and the scarcity of specialized knowledge. However, advancements in Artificial Intelligence (AI) offer promising solutions to these challenges. This review explores the current state of AI-powered diagnostic tools for rare diseases, focusing on the methodologies, algorithms, and platforms utilized in this emerging field. We examine how AI technologies, such as machine learning, deep learning, and natural language processing, are being integrated into clinical practice to enhance diagnostic accuracy and speed. The research also provides the examples that highlight the successes and limitations of AI in this domain, providing insights into how AI can be harnessed to improve patient outcomes in rare disease diagnosis and management.

ETLoViT: an acne diagnose approach using vision-transformers and model ensembling

Acne, a widespread skin condition predominantly affecting teenagers, presents intricate challenges in its diagnosis. Recent advances in deep learning, machine learning, and image processing methods have made it possible to diagnose acne automatically and effectively. However, achieving higher acne classification accuracy is still one of the concerns with these methods. Therefore, this paper introduces a group of trained models based on transfer learning that is applied to Vision Transformer extracted features (ETLoViT). These models are trained using two innovative deep-learning methods: the Vision Transformer (ViT) and model ensembling for acne image classification. The ViT approach harnesses the power of the attention module to extract acne features efficiently. These extracted features are subsequently run through different transfer learning models, such as MobileNetV2, VGG16, and InceptionResNetV2. The predicted results were subsequently combined for classification. The proposed model (ETLoViT) is compared with standing deep learning models, and the outcomes demonstrate that the proposed (ETLoViT) model consistently performs better, achieving an astounding 96% classification accuracy.

AI and Big Data approaches to addressing the opioid crisis: a scoping review protocol

IntroductionThis paper outlines the steps necessary to assess the latest developments in artificial intelligence (AI) as well as Big Data technologies and their relevance to the opioid crisis. Fatal opioid...

TinyBioGait—Embedded intelligence and homologous time approximation warping for gait biometric authentication from IMU signals

The gait of a subject follows a specific pattern, but variations exist that are unique to a subject but contrasting to other subjects. This can be utilized for biometric authentication to prevent impersonation during gait studies. However, due to the dynamic nature of gait, like changes in gait speed while walking, gait biometric authentications are challenging. In the state-of-the-art, although attempts have been made to use deep learning and other signal processing methods for biometric authentication, which obtained reliable results, these are either highly resource-consuming, require several sensors or need an expensive framework, making it challenging to implement this in many scenarios. Therefore, a knowledge gap exists to build a reliable, inexpensive and resource-efficient gait biometric authentication system. The paper proposes a method for using only one embedded IMU sensor with a microcontroller for tracking the motion of a subject, resource-efficient on-device elimination of the gait speed differences by proposing a homologous time approximation warping algorithm and building a resource-efficient TinyML model for reliable biometric authentication. Based on an experiment consisting of 20 human subjects with consent, the microcontroller’s on-device accuracy score for decision-making by TinyML was found to be 0.9276. The resource efficiency of the model based on memory profiling has been further discussed. Also, the prediction performance of the microcontroller with the proposed optimization was found to be only 8% slower compared to a personal computer, given that several thousands of processes run parallel on a personal computer. The work needs to be further tested for a larger sample space, and data privacy needs to be addressed.

ForestAdvisor: A multi-modal forest decision-making system based on carbon emissions

Effectively balancing carbon emission reduction with economic viability through regional forest management is a significant challenge for global ecosystems. This paper introduces an innovative multi-modal forest decision-making system, integrating deep learning and natural language processing technologies, aimed at optimizing forest management strategies. Experimental validation of this system was conducted in three distinct forested regions. Utilizing a deep learning model, the system analyzed and predicted daily carbon emissions data. The experiments demonstrated remarkable accuracy, with the model achieving a coefficient of determination (R2) of up to 0.94, 0.98, and 0.99 across datasets from all three regions, thereby justifying its use for forecasting carbon emission trends over the following months. Subsequently, the system employed natural language processing to assess the importance of various collected forest management strategies. Finally, the system fine-tuned these strategy combinations in response to the predicted carbon emission trends, ensuring flexibility and effectiveness in addressing the complex dynamics of carbon emission fluctuations.

Active Learning Pipeline to Identify Candidate Terms for a CDSS Ontology.

Ontology is essential for achieving health information and information technology application interoperability in the biomedical fields and beyond. Traditionally, ontology construction is carried out manually by human domain experts (HDE). Here, we explore an active learning approach to automatically identify candidate terms from publications, with manual verification later as a part of a deep learning model training and learning process. We introduce the overall architecture of the active learning pipeline and present some preliminary results. This work is a critical and complementary component in addition to manually building the ontology, especially during the long-term maintenance stage.

Enhancing Agile Story Point Estimation: Integrating Deep Learning, Machine Learning, and Natural Language Processing with SBERT and Gradient Boosted Trees

Advances in software engineering, particularly in Agile software development (ASD), demand innovative approaches to effort estimation due to the volatility in Agile environments. Recent trends have made the automation of story point (SP) estimation increasingly relevant, with significant potential for enhancing accuracy. This study introduces a novel model for software effort estimation (SEE) utilizing a deep learning (DL)-based sentence-BERT (SBERT) model for feature extraction combined with advanced gradient-boosted tree (GBT) algorithms. A comprehensive evaluation shows that the proposed model outperforms standard SEE and state-of-the-art models, demonstrating a mean absolute error (MAE) of 2.15 and a median absolute error (MdAE) of 1.85, representing a 12% improvement over the baseline model and an 18% improvement over the best-performing state-of-the-art model. The standardized accuracy (SA) is 93%, which is 7% higher than the next best model, across a dataset of 31,960 issues from 26 open-source Agile projects. This study contributes to software engineering by offering a more accurate and reliable decision support system for estimating project efforts.

Optimized design of hardware accelerator based on machine learning

Hardware accelerators are playing an increasingly important role in the field of computer science and engineering, especially in accelerating deep learning and large-scale data processing. However, the traditional hardware accelerator design often relies on manual adjustment and experience accumulation, which is inefficient. Therefore, the use of machine learning technology to optimize hardware accelerator design has become a new trend. In this paper, a hardware accelerator optimization design method based on machine learning is proposed, including hardware design, algorithm optimization, system-level optimization, and deep learning model compression. Through the research and verification of the actual case, this research proves the effectiveness and feasibility of the hardware accelerator optimization design method based on machine learning. Compared with the traditional design method, this method can find better design scheme in shorter time and has higher performance prediction accuracy. Therefore, this method can significantly improve the design efficiency and performance of hardware accelerators. The results of this study are of great significance for promoting the development of hardware accelerator technology. It can promote scientific research and engineering applications in related fields to make greater progress, and promote the development of artificial intelligence, big data processing and other fields.

Advancements in deep learning and natural language processing for effective disaster sentiment analysis: A review

This article comprehensively studies the application of deep learning (DL), natural language processing (NLP), and large language models (LLM) in sentiment analysis in disaster scenarios such as earthquakes and major accidents. The article focuses on the latest developments in these technologies and their role in strengthening disaster management and response. The article explores various methods, including BERT, LSTM, and convolutional neural networks, with a focus on their practicality, challenges, and potential for development. This review aims to provide researchers and relevant practitioners with a comprehensive understanding of this rapidly developing field.

Low-illumination and noisy bridge crack image restoration by deep CNN denoiser and normalized flow module

When applying deep learning and image processing techniques for bridge crack detection, the obtained images in real-world scenarios have severe image degradation problem. This study focuses on restoring low-illumination bridge crack images corrupted by noise to improve the accuracy of subsequent crack detection and semantic segmentation. The proposed algorithm consists of a deep CNN denoiser and a normalized flow-based brightness enhancement module. By taking the noise spectrum as an input, the deep CNN denoiser restores image at a broad range of noise levels. The normalized flow module, employs a conditional encoder and a reversible network to map the distribution of normally exposed images to a Gaussian distribution, effectively improving the image brightness. Extensive experiments have demonstrated the approach can usefully recover low-illumination images corrupted by noise compared to the state-of-the-art methods. Furthermore, the algorithm presented in this study can also be applied to other image quality restoration with high generalization and robust abilities. And the semantic segmentation accuracy of the restored image is significantly improved.

Deep Ensemble Learning-Based Sensor for Flotation Froth Image Recognition.

Froth flotation is a widespread and important method for mineral separation, significantly influencing the purity and quality of extracted minerals. Traditionally, workers need to control chemical dosages by observing the visual characteristics of flotation froth, but this requires considerable experience and operational skills. This paper designs a deep ensemble learning-based sensor for flotation froth image recognition to monitor actual flotation froth working conditions, so as to assist operators in facilitating chemical dosage adjustments and achieve the industrial goals of promoting concentrate grade and mineral recovery. In our approach, training and validation data on flotation froth images are partitioned in K-fold cross validation, and deep neural network (DNN) based learners are generated through pre-trained DNN models in image-enhanced training data, in order to improve their generalization and robustness. Then, a membership function utilizing the performance information of the DNN-based learners during the validation is proposed to improve the recognition accuracy of the DNN-based learners. Subsequently, a technique for order preference by similarity to an ideal solution (TOPSIS) based on the F1 score is proposed to select the most probable working condition of flotation froth images through a decision matrix composed of the DNN-based learners' predictions via a membership function, which is adopted to optimize the combination process of deep ensemble learning. The effectiveness and superiority of the designed sensor are verified in a real industrial gold-antimony froth flotation application.

The Role of Artificial Intelligence in Treatment and Diagnosis in Healthcare

Technology, specifically artificial intelligence (AI) is gradually but progressively creeping into the health sector and it’s perhaps the one that has been revolutionised most in diagnosis and treatment. This review brings out discussions on the practices of AI technologies in medical, the pros and the cons. First of all, an endeavour is made to elucidate the meaning of the term AI and its utilization in the field of healthcare. The specific AI techniques are described comprehensively focusing on the machine learning, deep learning, and natural language processing methods to be used in the project The role of multiple types of data in AI includes the EHR, medical images, and genomics data. Self-diagnosis: AI is improving the diagnosis approaches in the radiology and pathology fields and predicting the early-stage disease with better results in most of the cases, and enhancing the identification of genetic diseases. As for treatment, the enhancement of the use of AI has had an impact on issues such as; Prescribing and recommending drugs according to the characteristics of the patients, smart drug administration and management, robotic surgeries and simulations. Discussions are made using concrete and successful implementation of AI in cancer, cardiovascular, neurological and infectious diseases for the purpose of elucidating particular results. This also has to do with the ethical and legal problems like who has the liability to determine in the instance of complicated problems, patients’ information discretion, data privacy, and other legalities. In this article, we briefly mention the prosaic matters of AI, which deals with the engineering aspects of establishing AI such as the aspect of data and the ways and means of checking them and the interdisciplinary character of it. Concerning future developments, additional technologies like AI and connected devices in the field of health care, interdisciplinary at national and international level as well as data sharing is emphasized. Thus. AI has a very great perspective in healthcare, particularly in diagnostics and treatment of diseases due to the probability of increasing the level of accuracy, efficacy, and personalization. Despite these, they are tangible objectives with major challenges and require cooperation between nations with proper handling of Artificial Intelligence to practice clinical medication.

Deep Learning and Fusion Mechanism-based Multimodal Fake News Detection Methodologies: A Review

Today, detecting fake news has become challenging as anyone can interact by freely sending or receiving electronic information. Deep learning processes to detect multimodal fake news have achieved great success. However, these methods easily fuse information from different modality sources, such as concatenation and element-wise product, without considering how each modality affects the other, resulting in low accuracy. This study presents a focused survey on the use of deep learning approaches to detect multimodal visual and textual fake news on various social networks from 2019 to 2024. Several relevant factors are discussed, including a) the detection stage, which involves deep learning algorithms, b) methods for analyzing various data types, and c) choosing the best fusion mechanism to combine multiple data sources. This study delves into the existing constraints of previous studies to provide future tips for addressing open challenges and problems.

Explainable Detection of Fake News on Social Media Using Pyramidal Co-Attention Network

In today’s world, fake news on social media is a universal trend and has severe consequences. There has been a wide variety of countermeasures developed to offset the effect and propagation of Fake News. The most common are linguistic-based techniques, which mostly use deep learning (DL) and natural language processing (NLP). Even government-sponsored organizations spread fake news as a cyberwar strategy. In literature, computational-based detection of fake news has been investigated to minimize it. The initial results of these studies are good but not significant. However, we argue that the explainability of such detection, particularly why a certain news item is detected as fake, is a vital missing element of the studies. In real-world settings, the explainability of the system’s decisions is just as important as its accuracy. This article explores explainable fake news detection and proposes a sentence-comment-based co-attention sub-network model. The proposed model uses user comments and news contents to mutually apprehend top- <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$k$</tex-math> </inline-formula> explainable check-worthy user comments and sentences for detecting fake news. The experimental result on real-world datasets shows that our proposed model outperforms state-of-the-art techniques by 5.56% in the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$F$</tex-math> </inline-formula> 1 score. In addition, our model outperforms other baselines by 16.4% in normalized cumulative gain (NDCG) and 22.1% in Precision in identifying top- <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$k$</tex-math> </inline-formula> comments from users, which indicates why a news article can be fake.

Advancements, Applications, and Future Directions of Artificial Intelligence in Healthcare

Background: The integration of artificial intelligence (AI) into healthcare represents a transformative shift in medical procedures, offering substantial benefits across various domains. With advancements in AI technologies such as machine learning (ML), deep learning (DL), and natural language processing (NLP), healthcare systems are witnessing improvements in early detection, patient treatment, and overall administration. This article traces the evolution of AI, from foundational contributions by Alan Turing during World War II to contemporary applications like ChatGPT, and examines the impact of AI in enhancing diagnostic accuracy and treatment outcomes. Methods: This comprehensive review analyzes the existing literature on AI applications in healthcare, focusing on various AI methodologies and their integration into clinical settings. It evaluates the effectiveness of AI in processing large datasets, improving diagnostic precision, and facilitating data-driven decision-making. The study also explores the ethical, legal, and technical challenges associated with AI deployment in medical environments. Results: AI technologies have demonstrated significant improvements in healthcare, particularly in early disease detection, personalized treatment plans, and resource management. The use of AI in analyzing vast medical datasets has enhanced diagnostic accuracy, reduced costs, and optimized patient care. However, challenges related to ethical considerations, patient privacy, and system reliability remain critical barriers to full-scale AI adoption. Conclusion: Despite the challenges, AI is positioned as an indispensable tool in modern medicine, capable of enhancing preventive care, personalizing treatments, and improving healthcare delivery. This review proposes a framework for evaluating the benefits, challenges, and strategies of AI integration in healthcare. Further research is essential to maximize AI's potential while addressing ethical and practical concerns, ensuring safe and effective implementation in clinical settings.

Estimation of winter wheat yield from sentinel-1A time-series images using ensemble deep learning and a Gaussian process

ABSTRACT Deep learning methods have been widely used in applications of yield estimation using remote sensing data. However, they still face some challenges in terms of accuracy due to their inability to fully utilize the spatiotemporal information of remote sensing data. In response to this problem, a winter wheat yield estimation method based on ensemble deep learning and Gaussian process (GP) was proposed. First, the long-short term memory (LSTM) network and convolutional neural network (CNN) were constructed to explore deep spatiotemporal features from Sentinel-1A time-series images. Then, a GP component was applied for fusion of intraimage deep spatiotemporal features and inter-sample spatial consistency features. Finally, the yield estimation results were obtained. The experimental results showed that the proposed method had higher accuracy than those compared models, with a coefficient of determination ( R 2 ) of 0.698, a root mean square error (RMSE) of 477.045 kg/ha and a mean absolute error (MAE) of 404.377 kg/ha, demonstrating the application potential of the proposed method in crop yield estimation applications.

Enhanced Nanoparticle Recognition via Deep Learning-Accelerated Plasmonic Sensing.

Surface plasmon microscopy proves to be a potent tool for capturing interferometric scattering imaging data of individual particles at both micro and nanoscales, offering considerable potential for label-free analysis of bio-particles and bio-molecules such as exosomes, viruses, and bacteria. However, the manual analysis of acquired images remains a challenge, particularly when dealing with dense samples or strong background noise, common in practical measurements. Manual analysis is not only prone to errors but is also time-consuming, especially when handling a large volume of experimental images. Currently, automated methods for sensing and analysis of such data are lacking. In this paper, we develop an accelerated approach for surface plasmon microscopy imaging of individual particles based on combining the interference scattering model of single particle and deep learning processing. We create hybrid datasets by combining the theoretical simulation of particle images with the actual measurements. Subsequently, we construct a neural network utilizing the EfficientNet architecture. Our results demonstrate the effectiveness of this novel deep learning technique in classifying interferometric scattering images and identifying multiple particles under noisy conditions. This advancement paves the way for practical bio-applications through efficient automated particle analysis.

REDef-DETR: real-time and efficient DETR for industrial surface defect detection

Abstract Industrial surface defect detection is an important part of industrial production, which aims to identify and detecting various defects on the surface of product to ensure quality and meet customer requirements. With the development of deep learning and image processing technologies, the surface defect detection methods based on computer vision has become the mainstream method. However, the prevalent convolutional neural network-based defect detection methods also have many problems. For example, these methods rely on post-processing of Non-Maximum Suppression and have poor detection ability for small targets, which affects the speed and accuracy of surface defect detection in industrial scenarios. Therefore, we propose a novel DEtection TRansformer-based surface defect detection method. Firstly, we propose a Multi-scale Contextual Information Dilated module and fuse it into the backbone. The module is mainly composed of large kernel convolutions, which aims to expand the receptive field of the model, thus reducing the leakage rate of the model. Moreover, we design an efficient encoder which mainly contains two important modules, namely feature enhancement based on cascaded group attention module and efficient feature fusion module based on content-aware. The former module effectively enhances the high-level semantic information extracted by the backbone, thus enabling the model to better interpret features, and it can improve the problem of high computational cost of transformer encoder, thus increasing the detection speed. The latter module performs multi-scale feature fusion across the feature information of various scales, thus improving the detection accuracy of the model for small-size defects. Experimental results show that the proposed method achieves 80.6%mAP and 80.3FPS on NEU-DET, and 98.0%mAP and 79.4FPS on PCB-DET. Our proposed method exhibits excellent detection performance and achieves real-time and efficient surface defect detection capability to meet the needs of industrial surface defect detection.

Classification of food items in trash through images using a hybrid deep learning system

The issue that arouses concern is the issue of waste in general, especially waste in food. Deep learning has made its way into the aforementioned field. Pre-processing is done using a new filter derived from discrete wavelets that were derived from Legendre polynomials (LEGP) to achieve Multi Resolution Analyses (MRA), in which the image parameters are separated into proximity parameters and detail parameters, where the first is concentrated in LL, while the second is in HL, LH, HH to improve the input image of the waste in terms of noise removal and image compression. To begin the deep learning process, configure the New Convolutional Neural Network (NCNN) to configure 144 layer with the image size (224x224x3) with Google Net so that the waste resulting from leftover food is sorted. In this work, four foods were sorted: tomatoes, potatoes, bread, and rice, and the result was reached for accuracy for each upcoming image with the accuracy of the neural arrangement of rice (99.9337) to potato (100 %) to tomato (99.8959), and finally for bread (83.5492), with the accuracy of the neural network 94.12 % in 1 min 2 sec. The accuracy reached in this work with Discrete Legendre Wavelets Transform (DLEGWT) and training the new network for google net was 94.12 % in 1 minute and 2 seconds with the new filter (DLEGWT). A program was created that was translated into a new algorithm. It was shown that the new filter worked in two stages, which is the initial processing stage, included analyses image with removed the noise from the image, after which the image is compressed to reduce the space used by the image data. As for the second stage, the deep learning stage takes place with the new filter, so that a new convolutional neural network is created after Dividing the image into three channels RGB to complete the convolution process for each channel, so that the hidden layers are formed until it reaches full connectivity, so that the classification process is performed with triple accuracy. The proposed algorithm has proven its efficiency