New Insights in Machine Learning (ML) and Deep Neural Networks

A special issue of Mathematics (ISSN 2227-7390). This special issue belongs to the section "E1: Mathematics and Computer Science".

Deadline for manuscript submissions: 31 January 2026 | Viewed by 6923

Special Issue Editors


E-Mail Website
Guest Editor
College of Systems Engineering, National University of Defense Technology, Changsha 410073, China
Interests: image retrieval; computer vision; image processing; pattern recognition; feature extraction; machine learning; object recognition; classification; algorithms; image data analysis; machine vision; image recognition; pattern classification; visual pattern recognition; multimedia analysis

E-Mail Website
Guest Editor
School of Software Technology, DUT-RU International School of Information Science and Engineering, Dalian University of Technology, Dalian 116024, China
Interests: open-world vision perception; continual/incremental learning; multimedia retrieval; object detection/segmentation

E-Mail Website
Guest Editor
College of Systems Engineering, National University of Defense Technology, Changsha 410072, China
Interests: artificial intelligence models; convolutional neural networks; decision tree; deep learning

E-Mail Website
Guest Editor Assistant
Laboratory for Big Data and Decision, College of Systems Engineering, National University of Defense Technology, Changsha 410073, China
Interests: multimodal information comprehension; visual question answering; machine reading comprehension; deep learning theories

Special Issue Information

Dear Colleagues,

The confluence of groundbreaking research and technological innovation in the realm of Machine Learning (ML) and Deep Neural Networks (DNNs) continues to redefine the horizons of artificial intelligence. As we stand on the precipice of new discoveries that have the potential to revolutionize various sectors, this is the time to explore and document these advancements. With this in mind, we are proud to announce a Special Issue themed "New Insights in Machine Learning (ML) and Deep Neural Networks", aimed at capturing the essence of current and emerging trends in the field.

This Special Issue is a dedicated forum for scholars and professionals to share their latest findings, innovative methodologies, and transformative applications of ML and DNNs. We are particularly interested in works that address, but are not limited to, the following areas of exploration:

  • Parameter Efficient Fine-Tuning Methods of Large Language Models (LLMs);
  • Trustworthy Machine Learning;
  • Adversarial Attacks and Defenses;
  • Generative Multi-Modal Models;
  • Continual Learning;
  • Embodied Intelligence;
  • AI for Science;
  • Multimodal Understanding and Generation.

We also welcome review articles that synthesize the current state of the art, providing comprehensive overviews and identifying future research directions in the areas pertinent to this Special Issue. 

Dr. Yanming Guo
Dr. Yu Liu
Dr. Tianyuan Yu
Guest Editors

Dr. Mingrui Lao
Guest Editor Assistant

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Mathematics is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2600 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

Keywords

  • machine learning
  • multi-modal models
  • embodied intelligence
  • artificial intelligence
  • deep neural networks (DNNs)

Benefits of Publishing in a Special Issue

  • Ease of navigation: Grouping papers by topic helps scholars navigate broad scope journals more efficiently.
  • Greater discoverability: Special Issues support the reach and impact of scientific research. Articles in Special Issues are more discoverable and cited more frequently.
  • Expansion of research network: Special Issues facilitate connections among authors, fostering scientific collaborations.
  • External promotion: Articles in Special Issues are often promoted through the journal's social media, increasing their visibility.
  • Reprint: MDPI Books provides the opportunity to republish successful Special Issues in book format, both online and in print.

Further information on MDPI's Special Issue policies can be found here.

Published Papers (5 papers)

Order results
Result details
Select all
Export citation of selected articles as:

Research

19 pages, 929 KiB  
Article
Online Banking Fraud Detection Model: Decentralized Machine Learning Framework to Enhance Effectiveness and Compliance with Data Privacy Regulations
by Hisham AbouGrad and Lakshmi Sankuru
Mathematics 2025, 13(13), 2110; https://doi.org/10.3390/math13132110 - 27 Jun 2025
Viewed by 71
Abstract
In such a dynamic and increasingly digitalized financial sector, many sophisticated fraudulent and cybercrime activities continue to challenge conventional detection systems. This research study explores a decentralized anomaly detection framework using deep autoencoders, designed to meet the dual imperatives of fraud detection effectiveness [...] Read more.
In such a dynamic and increasingly digitalized financial sector, many sophisticated fraudulent and cybercrime activities continue to challenge conventional detection systems. This research study explores a decentralized anomaly detection framework using deep autoencoders, designed to meet the dual imperatives of fraud detection effectiveness and user data privacy. Instead of relying on centralized aggregation or data sharing, the proposed model simulates distributed training across multiple financial nodes, with each institution processing data locally and independently. The framework is evaluated using two real-world datasets, the Credit Card Fraud dataset and the NeurIPS 2022 Bank Account Fraud dataset. The research methodology applied robust preprocessing, the implementation of a compact autoencoder architecture, and a threshold-based anomaly detection strategy. Evaluation metrics, such as confusion matrices, receiver operating characteristic (ROC) curves, precision–recall (PR) curves, and reconstruction error distributions, are used to assess the model’s performance. Also, a threshold sensitivity analysis has been applied to explore detection trade-offs at varying levels of strictness. Although the model’s recall remains modest due to class imbalance, it demonstrates strong precision at higher thresholds, which demonstrates its utility in minimizing false positives. Overall, this research study is a practical and privacy-conscious approach to fraud detection that aligns with the operational realities of financial institutions and regulatory compliance toward scalability, privacy preservation, and interpretable fraud detection solutions suitable for real-world financial environments. Full article
(This article belongs to the Special Issue New Insights in Machine Learning (ML) and Deep Neural Networks)
Show Figures

Figure 1

18 pages, 2831 KiB  
Article
Self-Supervised Feature Disentanglement for Deepfake Detection
by Bo Yan, Pan Liu, Yumin Yang and Yanming Guo
Mathematics 2025, 13(12), 2024; https://doi.org/10.3390/math13122024 - 19 Jun 2025
Viewed by 702
Abstract
Existing deepfake detection methods heavily rely on specific training data distributions and struggle to generalize to unknown forgery techniques. To address the challenge, this paper focuses on two critical research gaps: (1) the lack of systematic mining of standard features across multiple forgery [...] Read more.
Existing deepfake detection methods heavily rely on specific training data distributions and struggle to generalize to unknown forgery techniques. To address the challenge, this paper focuses on two critical research gaps: (1) the lack of systematic mining of standard features across multiple forgery methods; (2) the unresolved distribution shift problem in the strong supervised learning paradigm. To tackle these issues, we propose a self-supervised learning framework based on feature disentanglement, which enhances the generalization ability of detection models by uncovering the intrinsic features of forged content. The core method comprises three key components: self-supervised sample construction and training samples for feature disentanglement, which are generated via an image self-mixing mechanism; feature disentanglement network, where the input image is decomposed into two parts—content features irrelevant to forgery and discriminative forgery-related features; and conditional decoder verification, where both types of features are used to reconstruct the image, with forgery-related features serving as conditional vectors to guide the reconstruction process. Orthogonal constraints on features are enforced to mitigate the overfitting problem in traditional methods. Experimental results demonstrate that, compared with state-of-the-art methods, the proposed framework exhibits superior generalization performance in cross-unknown forgery technique detection tasks, effectively breaking through the dependency bottleneck of traditional supervised learning on training data distributions. This study provides a universal solution for deepfake detection that does not rely on specific forgery techniques. The model’s robustness in real-world complex scenarios is significantly improved by mining the common essence of forgery features. Full article
(This article belongs to the Special Issue New Insights in Machine Learning (ML) and Deep Neural Networks)
Show Figures

Figure 1

16 pages, 503 KiB  
Article
Overcoming Class Imbalance in Incremental Learning Using an Elastic Weight Consolidation-Assisted Common Encoder Approach
by Engin Baysal and Cüneyt Bayılmış
Mathematics 2025, 13(11), 1887; https://doi.org/10.3390/math13111887 - 4 Jun 2025
Viewed by 537
Abstract
Incremental learning empowers models to continuously acquire knowledge of new classes while retaining previously learned information. However, catastrophic forgetting and class imbalance often impede this process, especially when new classes are introduced sequentially. We propose a hybrid method that integrates Elastic Weight Consolidation [...] Read more.
Incremental learning empowers models to continuously acquire knowledge of new classes while retaining previously learned information. However, catastrophic forgetting and class imbalance often impede this process, especially when new classes are introduced sequentially. We propose a hybrid method that integrates Elastic Weight Consolidation (EWC) with a shared encoder architecture to overcome these obstacles. This approach provides robust feature extraction, while EWC safeguards vital parameters and preserves prior knowledge. Moreover, task-specific output layers enable flexible adaptation to new classes. We evaluated our method using the CICIoT2023 dataset, a class-incremental IoT anomaly detection benchmark. Our results demonstrated a 15.3% improvement in the macro F1-score and a 1.28% increase in overall accuracy compared to a baseline model that did not incorporate EWC, with particular advantages for underrepresented classes. These findings underscore the effectiveness of the EWC-assisted shared encoder framework for class-imbalanced incremental learning in streaming environments. Full article
(This article belongs to the Special Issue New Insights in Machine Learning (ML) and Deep Neural Networks)
Show Figures

Figure 1

19 pages, 1572 KiB  
Article
FeTT: Class-Incremental Learning with Feature Transformation Tuning
by Sunyuan Qiang and Yanyan Liang
Mathematics 2025, 13(7), 1095; https://doi.org/10.3390/math13071095 - 27 Mar 2025
Viewed by 518
Abstract
Class-incremental learning (CIL) enables models to continuously acquire knowledge and adapt in an ever-changing environment. However, one primary challenge lies in the trade-off between the stability and plasticity, i.e., plastically expand the novel knowledge base and stably retaining previous knowledge without catastrophic forgetting. [...] Read more.
Class-incremental learning (CIL) enables models to continuously acquire knowledge and adapt in an ever-changing environment. However, one primary challenge lies in the trade-off between the stability and plasticity, i.e., plastically expand the novel knowledge base and stably retaining previous knowledge without catastrophic forgetting. We find that even recent promising CIL methods via pre-trained models (PTMs) still suffer from this dilemma. To this end, this paper begins by analyzing the aforementioned dilemma from the perspective of marginal distribution for data categories. Then, we propose the feature transformation tuning (FeTT) model, which concurrently alleviates the inadequacy of previous PTM-based CIL in terms of stability and plasticity. Specifically, we apply the parameter-efficient fine-tuning (PEFT) strategies solely in the first CIL task to bridge the domain gap between the PTMs and downstream task dataset. Subsequently, the model is kept fixed to maintain stability and avoid discrepancies in training data distributions. Moreover, feature transformation is employed to regulate the backbone representations, boosting the model’s adaptability and plasticity without additional training or parameter costs. Extensive experimental results and further feature channel activations discussion on CIL benchmarks across six datasets validate the superior performance of our proposed method. Full article
(This article belongs to the Special Issue New Insights in Machine Learning (ML) and Deep Neural Networks)
Show Figures

Figure 1

18 pages, 393 KiB  
Article
LLM-Augmented Linear Transformer–CNN for Enhanced Stock Price Prediction
by Lei Zhou, Yuqi Zhang, Jian Yu, Guiling Wang, Zhizhong Liu, Sira Yongchareon and Nancy Wang
Mathematics 2025, 13(3), 487; https://doi.org/10.3390/math13030487 - 31 Jan 2025
Cited by 5 | Viewed by 4436
Abstract
Accurately predicting stock prices remains a challenging task due to the volatile and complex nature of financial markets. In this study, we propose a novel hybrid deep learning framework that integrates a large language model (LLM), a Linear Transformer (LT), and a Convolutional [...] Read more.
Accurately predicting stock prices remains a challenging task due to the volatile and complex nature of financial markets. In this study, we propose a novel hybrid deep learning framework that integrates a large language model (LLM), a Linear Transformer (LT), and a Convolutional Neural Network (CNN) to enhance stock price prediction using solely historical market data. The framework leverages the LLM as a professional financial analyst to perform daily technical analysis. The technical indicators, including moving averages (MAs), relative strength index (RSI), and Bollinger Bands (BBs), are calculated directly from historical stock data. These indicators are then analyzed by the LLM, generating descriptive textual summaries. The textual summaries are further transformed into vector representations using FinBERT, a pre-trained financial language model, to enhance the dataset with contextual insights. The FinBERT embeddings are integrated with features from two additional branches: the Linear Transformer branch, which captures long-term dependencies in time-series stock data through a linearized self-attention mechanism, and the CNN branch, which extracts spatial features from visual representations of stock chart data. The combined features from these three modalities are then processed by a Feedforward Neural Network (FNN) for final stock price prediction. Experimental results on the S&P 500 dataset demonstrate that the proposed framework significantly improves stock prediction accuracy by effectively capturing temporal, spatial, and contextual dependencies in the data. This multimodal approach highlights the importance of integrating advanced technical analysis with deep learning architectures for enhanced financial forecasting. Full article
(This article belongs to the Special Issue New Insights in Machine Learning (ML) and Deep Neural Networks)
Show Figures

Figure 1

Back to TopTop