Journal of Applied Artificial Intelligence

Ensemble-based Intrusion Detection System for Electric Vehicles Charging Stations using Machine Learning

Bishal K C, Kshitiz Aryal, Sansrit Paudel — Wed, 31 Dec 2025 00:00:00 +0000

Traditional Vehicles have an adverse effect on the environment. Therefore, the current technological shift is constantly seeking an alternative to replace traditional vehicles fueled by fossil fuels, and Electric vehicles are, so far, the best alternative. The adoption of Electric Vehicles (EVs) is growing rapidly due to their eco-friendly benefits and technological advancements. This growth, however, brings a significantly larger attack surface due to increased interconnectivity between electric vehicles, charging stations and the smart grid system. To prevent such types of attacks, we need a robust system to detect them beforehand and prevent the system from being compromised. Although some prior work has been conducted in this area, their approaches did not incorporate deep learning algorithms, nor did they evaluate model performance under noisy data conditions. Therefore, we proposed a novel ensemble-based intrusion detection system (IDS) to detect these attacks in Electric Vehicle Charging Stations (EVCS). We implement different Machine learning algorithms such as k-nearest neighbors (KNN), Logistic Regression (LR), Support Vector Machine (SVM) and Decision Tree (DT). Moreover, as different types of malwares often exhibit distinct structural characteristics when visualized as images, we also use Convolutional Neural Networks (CNNs) to detect such attacks and malware. We are focusing on detecting attacks in Electric vehicle charging stations by analyzing the network traffic. For this, we utilize the latest labelled dataset, the Canadian Institute of Cybersecurity EV Charger Attack Dataset 2024 (CICEVSE2024), which is a multidimensional dataset containing both benign and attack data. We then evaluate & compare the performance of these algorithm in detecting the network traffic attacks in Electric Vehicle Charging Stations (EVCS). Our proposed model employs an ensemble voting strategy to combine the predictions from different classifiers, thereby improving the system's robustness and accuracy, and achieves an accuracy of 99.5% in detecting cyberattacks. With the addition of small noise to the dataset, a few individual classifiers perform poorly; however, the ensemble model still maintains an accuracy of 99.2%.

DCGAN Beyond Generation: A Critical Review of The Performance and Challenges of Forensic Face Models

HASAN AL-MUTTAIRI; Mr. Mahmood — Wed, 31 Dec 2025 00:00:00 +0000

Computer vision and deep learning techniques, especially deep convolutional generative adversarial networks (DCGAN), have enabled advanced mechanisms to address complex challenges in forensics, especially in reactivating cold case investigations. Cold cases present unresolved challenges due to deteriorating or scarce visual evidence. This paper provides a systematic review that analyzes, classifies, and evaluates the current status of DCGAN and related GAN structures in legitimate face modeling. The primary objectives are to evaluate reported methodologies, performance metrics, and limitations across key applications, including sketch-to-image conversion. The review identifies significant methodological gaps, particularly the absence of standardized assessment measures and the critical challenge of identity preservation. Furthermore, the research explores the ethical and legal considerations associated with computer-generated facial images, focusing on algorithmic bias, accountability, and legal admissibility in criminal investigations. The paper concludes by highlighting key research gaps and proposing future directions necessary for the robust, reliable, and ethically responsible deployment of GAN systems in legitimate practice

A Bi-Model Machine Learning Driven Application for Diagnosing the Dominant Illness among Typical Nigerian University Students

Dauda Olorunkemi Isiaka, Malik Adeiza Rufai , Abubakar Aliyu — Wed, 31 Dec 2025 00:00:00 +0000

The implementation and deployment of machine learning models for the diagnosis of dominant illnesses among students require significant investment in technology and infrastructure, which is among the barriers for healthcare organizations with limited resources. In order to increase its adoption, this research suggests the creation of a Bi-Model Machine Learning Driven Application that will enable university students to get diagnosed with common ailments. The plan is to apply a high-level model using a hybrid methodology that combines the development of Machine Learning Models with Agile Software Development. In order to do this specifically, Python was used to implement exploratory data analysis, classification, and regression models, as they have proven to be highly effective in both diagnosing the primary illness and predicting the length of hospital stay. The bi-model were built with four different algorithms each, so as to adopt the ones with best performance for the deployment. The model built with Gradient Boosting Classifier has 100% accuracy, 100% precision, 100% recall as compared to other three algorithms through three repeated training of the model. On the prediction of admission duration task, Gradient Boost Regression works best, and this is because it has the least Root Mean Square Error of 0.57 and Mean Absolute Error to be 0.423 among other compared three algorithms, as measured. This was achieved through the use of fresh localized dataset from the Federal University Lokoja Health Center, which was pre-processed, and stored in the file manager/internal storage for visualization and modelling. Furthermore, the completed models was deployed to a web application using flask and Mysql Lite Database. In the end, the application reduced human error in diagnosis and care management of the student population while they are pursuing their education by enabling evidence-based awareness, educated public health policy, and individualized treatment.

Asynchronous Deep Reinforcement Learning: A Shared Experience Replay Framework

Luckyson Khaidem, Kai Xi — Wed, 31 Dec 2025 00:00:00 +0000

Off-policy reinforcement learning (RL) algorithms with experience replay have achieved strong performance across a range of decision-making tasks. However, traditional implementations typically rely on a single agent interacting with one environment instance, which can limit exploration diversity and slow convergence. In this paper, we propose an asynchronous multi-agent RL framework that leverages a shared experience replay buffer. Each agent interacts independently with its own environment instance, contributing to a centralized buffer that aggregates diverse trajectories. This setup enhances sample diversity, accelerates learning, and scales efficiently with modern hardware. Our framework is compatible with standard off-policy algorithms such as Double DQN (DDQN) and DDPG, and we demonstrate its effectiveness across discrete and continuous control benchmarks. Experimental results show that our approach significantly improves convergence speed and learning stability compared to single-agent baselines. We discuss the theoretical implications of sharng experiences across agents and highlight real-world applications and future extensions, including hierarchical coordination, prioritized sampling, and deployment in real-time control systems.

Privacy-Preserving Artificial Intelligence: Principles, Methods, Applications, and Challenges

OLAYEMI OLASEHINDE, Boniface Kayode Alese, Ojonukpe Eqwuche — Wed, 31 Dec 2025 00:00:00 +0000

Artificial intelligence systems increasingly rely on large volumes of sensitive data to support decision making in domains such as healthcare, finance, education, and public administration. While these systems offer substantial benefits, their growing dependence on personal information has intensified concerns about privacy, data misuse, and loss of public trust. This study examines privacy-preserving artificial intelligence as a design approach that enables meaningful data analysis while limiting exposure of sensitive information. The paper analyses key privacy-preserving techniques, including federated learning, differential privacy, homomorphic encryption, and secure multi party computation, and evaluates their relevance across critical application domains. It also identifies practical challenges related to computational cost, data heterogeneity, regulatory compliance, and explainability. The study shows that privacy-preserving methods can support responsible and trustworthy artificial intelligence when privacy, utility, and governance considerations are addressed together.