Search
87 articles for “representation learning”
-
Deep Learning Approaches for Graph Similarity Computation: A Review
Abstract: Graph-structured data is increasingly prevalent in many real-world applications such as social networks, biological systems, molecular chemistry, and knowledge graphs. Measuring the similarity between graphs is a fundamental task in graph analytics, enabling applications including pattern recognition, recommendation systems, network alignment, and drug discovery. Traditional approaches for graph similarity computation, such as Graph Edit Distance and graph kernels, provide effective structural comparison but often suffer from high computational complexity and …
Published in Journal of Computer Technology & Applications · Vol. 17, Issue 2, 2026 Read article
-
Improving Reliability of Deep Learning System Under Distribution Shift Via Feature-Space Out-of-Distribution Detection
Abstract: Deep learning systems achieve strong predictive performance under independently and identically distributed conditions; however, their reliability can decline significantly when exposed to distribution shifts during deployment. This decline is often accompanied by overconfident predictions, which can introduce serious risks in safety-critical applications. In this work, we explore whether explicitly modeling the structure of the latent feature space can improve reliability through more effective out-ofdistribution detection. A statistically grounded framework based …
Published in International Journal of Bioinformatics and Computational Biology · Vol. 4, Issue 2, 2026 · pp. 66–76 Read article
-
Hybrid Approach for Community Detection Using Deep Learning Techniques
Abstract: Community detection in complex networks is a fundamental problem with applications across diverse domains, ranging from social networks to biological systems and beyond. Traditional methods based on graph theory have been widely used for identifying communities within networks. However, the intricate and evolving nature of modern networks demands more sophisticated approaches. This research work proposes a hybrid approach that combines the strengths of deep learning techniques with traditional community detection …
Published in Recent Trends in Electronics Communication Systems · Vol. 11, Issue 3, 2024 · pp. 18–26 Read article
-
Accelerating Unsupervised Feature Learning: Parallelized Training of Denoising Autoencoders
Abstract: Unsupervised representation learning has become a cornerstone of contemporary machine learning, enabling algorithms to extract informative features from un-labelled, high-dimensional data. This work investigates the efficacy of stacked denoising autoencoders (SDAEs) trained via parallelized stochastic gradient descent (SGD) as a scalable approach to feature extraction. By strategically leveraging multi-threaded computation, our study systematically examines the trade-offs between increased parallelism, training efficiency, and the preservation of model accuracy. Experiments on the …
Published in Journal of Image Processing & Pattern Recognition Progress · Vol. 12, Issue 3, 2025 · pp. 56–64 Read article
-
Self-Supervised Learning for Early Disease Detection in Low-Label Medical Image Datasets
Abstract: Getting reliable labeled data for medical image analysis is genuinely difficult. Radiologists, pathologists, and other specialist clinicians are expensive, slow, and not universally available — particularly in low-resource clinical environments across developing nations. Unlike natural image benchmarks such as ImageNet, which provide millions of meticulously labeled training examples, most publicly accessible medical imaging repositories offer only a few hundred to a few thousand annotated scans. Supervised deep learning models trained …
Published in Journal of Image Processing & Pattern Recognition Progress · Vol. 13, Issue 3, 2026 Read article
-
Advances in Deep Learning for Medical Image Analysis in the Era of Precision Medicine
Abstract: Medical imaging is fundamental to modern healthcare but analyzing the high-dimensional data requires advanced techniques. Manual image interpretation is time-consuming, subjective and limited in detecting complex patterns and minute details. Recent breakthroughs in Deep Learning offer transformative advances for unlocking clinically relevant information from medical images. This paper provides a comprehensive 6000+ word review of the current state-of-the-art Deep Learning techniques for medical image analysis including detailed coverage of key …
Published in Research and Reviews : Journal of Computational Biology · Vol. 12, Issue 2, 2023 · pp. 10–23 Read article
-
Using Convolutional Neural Networks (CNN) for Age and Gender Prediction
Abstract: The network, security, and care have all become more dependent on age and gender identification. It's commonly used for children's access to age-appropriate content. To expand its reach, social media uses it to provide layered adverts and marketing. Face recognition has progressed to the point where we need to map it out further in order to achieve more usable results using various methodologies. In this study, we suggest using deep …
Published in Journal of Instrumentation Technology & Innovations · Vol. 12, Issue 1, 2022 · pp. 27–32 Read article
-
Emerging Trends in Data Structures for Modern Machine Learning Applications
Abstract: In the realm of machine learning, data structures play a pivotal role in facilitating efficient data manipulation, storage, and retrieval, thereby significantly impacting the performance and scalability of machine learning algorithms. In recent years, the field of machine learning has witnessed the emergence of novel data structures tailored to address scalability and efficiency challenges inherent in handling large-scale and high-dimensional data. This study provides a look at the data preprocessing, …
Published in International Journal of Data Structure Studies · Vol. 2, Issue 1, 2024 · pp. 1–7 Read article
-
CNN-Based Diagnosis of Skin Cancer from Dermoscopic Images
Abstract: Skin cancer has become one of the diseases widely spread over the globe, with melanoma becoming a severe threat to one’s health. Detection of such diseases at the initial stage saves an individual from drastic damage. Using a Convolutional Neural Network (CNN) for detecting skin cancer through image classification as benign or malignant provides significant support to dermatological practice and reduces dependence solely on subjective visual examination. Dermatologists often face …
Published in Research and Reviews: Journal of Oncology and Hematology · Vol. 15, Issue 1, 2026 · pp. 37–42 Read article
-
Topology and Geometry in Data Science: Persistent Homology and Beyond
Abstract: In recent years, the interplay between topology, geometry, and data science has gained substantial momentum, offering powerful frameworks to analyze and interpret complex datasets. Traditional statistical and machine learning methods often rely on linear or metric- based assumptions, which may fail to capture the intrinsic structure of high-dimensional or nonlinear data. In contrast, topological and geometric methods provide shape-oriented, scale- invariant tools that focus on the continuity, connectivity, and global …
Published in Recent Trends in Mathematics · Vol. 1, Issue 2, 2024 · pp. 21–27 Read article
-
Unleashing the Power of Mathematics in Natural Language Processing: Sentiment Analysis Perspectives
Abstract: Sentiment analysis (SA), which utilizes natural language processing (NLP), computational linguistics, text analysis, image processing, and video processing to extract and analyze subjective information from the internet, social media, and other sources, is becoming increasingly popular in both the business world and the scientific community. It is even possible to model it so that it focuses on polarity, sentiments and emotions, urgency, and even goals. It is able to distinguish …
Published in Recent Trends in Programming languages · Vol. 10, Issue 2, 2023 · pp. 1–9 Read article
-
Quantum-Inspired Neural Networks: Accelerating AI for Large-Scale Data Processing
Abstract: Recently, the world of artificial intelligence has been buzzing with exciting ideas inspired by quantum computing, especially when it comes to processing large amounts of data. Introducing the Quantum-Inspired Neural Network (QINN), a novel approach to conventional neural networks that blends concepts from quantum mechanics with machine learning techniques. Unlike typical networks that rely on neurons, QINNs utilize qubit-based representations, enabling them to perform computations in a more flexible and …
Published in Journal of Software Engineering Tools & Technology Trends · Vol. 12, Issue 3, 2025 · pp. 12–17 Read article
-
Machine Learning Approach to Detect and Analyze Attention-Deficit/Hyperactivity Disorder
Abstract: Attention-Deficit/Hyperactivity Disorder (ADHD) is a neurodevelopmental disorder characterized by difficulties with attention, impulse control, behavioral regulation, and daily functioning that persist across childhood and adulthood. Clinical diagnosis is predominantly based on behavioral assessments and expert interpretation, which may result in subjectivity and delayed clinical decisions. To reduce reliance on subjective evaluation, this study introduces an automated ADHD identification framework that integrates resting-state functional Magnetic Resonance Imaging (rs-fMRI) with advanced machine …
Published in Research and Reviews: A Journal of Neuroscience · Vol. 16, Issue 2, 2026 · pp. 22–26 Read article
-
Text to Image Using Machine Learning
Abstract: In the era of digital transformation, our project addresses the convergence of computer vision and natural language processing to enhance user interaction and visual content creation. This project comprises three distinct modules: user authentication and session management, image colorization from grayscale inputs, and text-to-image generation. The login registration module provides secure access to the system, ensuring user privacy and data integrity. Once authenticated, users can utilize advanced computer vision techniques …
Published in Journal of Image Processing & Pattern Recognition Progress · Vol. 11, Issue 2, 2024 · pp. 35–41 Read article
-
Cyber Security Implementation Through Deep Learning Based Fraudulent URL Detection
Abstract: A malicious URL, sometimes known as Spam, malicious ads, and drive-by vulnerabilities may all be found on a site that is malevolent. To find rogue URLs, you need to be quick. Researchers previously used blacklisting, regular expression searching and signature matching in their investigations. These strategies are worthless for discovering new malicious URLs or updated versions of previously problematic URLs. A machine learning-based approach may be developed to overcome this …
Published in Recent Trends in Programming languages · Vol. 9, Issue 2, 2022 · pp. 35–42 Read article
-
Intrinsic Evaluation of Graph Embeddings: Assessing Clustering and Community Detection Performance
Abstract: This paper presents an intrinsic evaluation of some graph embedding techniques on clustering and community detection tasks. We analyze a diverse set of embedding methods, ranging from traditional techniques such as Laplacian eigenmaps to more recent approaches like graph autoencoders, high-order proximity preserved embedding (HOPE), and graph attention network (GAT), using two widely studied datasets, Cora and CiteSeer. Our evaluation relies on two main metrics: Silhouette score with respect to …
Published in International Journal of Algorithms Design and Analysis Review · Vol. 3, Issue 1, 2025 · pp. 40–48 Read article
-
Advancements in Battery Storage Technology for Renewable Energy Systems: Improving Reliability and Efficiency of Sustainable Energy
Abstract: Advances in battery storage technology are critical to improved reliability and efficiency of renewable energy systems, underpinning a sustainable energy future. Innovations exist in many different kinds of battery technologies- being developed and tested, and the leading contenders include innovations such as lithium-ion and sodium-ion, and much newer entrants like the solid-state and lithium-sulfur batteries. These developments respond to growing needs for sustainable solutions towards better integration of intermittent renewable …
Published in Journal of Alternate Energy Sources & Technologies · Vol. 17, Issue 2, 2026 Read article
-
Image-Based Crack Morphology Characterisation for Electrical Failure Analysis in Conductive Polymer Composites
Abstract: Electrical performance in conductive polymer composites is strongly governed by crack-network evolution, yet failure analysis typically relies on qualitative image inspection or electrical anomaly detection in isolation. This work proposes an end-to-end framework that converts optical/SEM crack imagery into a standardised crack morphology signature and quantitatively links it to electrical degradation indicators. A two-stage learning strategy is adopted: crack-representation pretraining using the public Concrete Crack Images for Classification dataset, followed …
Published in Journal of Polymer & Composites · Vol. 14, Issue 2, 2026 · pp. 1375-1386 Read article
-
A Comparative Study Between GSM and CNN to Develop Gesture Detection Based Alert System for Women Safety
Abstract: Women’s safety is a pressing issue in today’s world, and technology can play a crucial role in addressing it. This project introduces a facial expression recognition device that uses Convolution Neural Network (CNN) technology and develop it’s comparison with an expression system with use of GSM is done. Unlike traditional methods relying on manual activation or dedicated devices, this system reacts instantly to threatening situations by recognizing predefined gestures, ensuring …
Published in International Journal of Electrical Power and Machine Systems · Vol. 2, Issue 1, 2024 · pp. 24–30 Read article
-
Digital Twin and Machine Learning Framework for Predictive Maintenance of Nuclear Power Systems
Abstract: Nuclear power systems contain safety-critical components whose degradation can affect availability, reliability, and maintenance planning. Conventional maintenance strategies are largely based on scheduled inspection, threshold monitoring, and condition-based assessment, which may not fully exploit the large volume of continuously generated plant data. Digital Twin (DT) technology provides a mechanism for maintaining a synchronized virtual representation of a physical nuclear system, while machine learning (ML) can extract degradation patterns from multivariate …
Published in Journal of Nuclear Engineering & Technology · Vol. 16, Issue 2, 2026 Read article