International Journal of Computing and Related Technologies

A Comprehensive Review of Machine Learning Algorithms and Datasets for Robust Face Recognition

Fri, 31 Jul 2026 00:00:00 +0000

Face recognition has been established as an essential biometric technology in security, surveillance, and human-computer interaction systems for verifying identities. Despite blistering development, the current literature is frequently unable to provide a cohesive comparative study that integrates classical machine learning approaches, recent deep learning models, benchmark datasets, and evaluation procedures within a unified framework. This paper shows a qualitative and analytical review of the face recognition methods, including the entire pipeline, such as face detector and alignment, feature extraction, feature classification, and recognition with deep learning. The gap of the research covered by the present study consists in the fact that a unified assessment, quantitatively comparing the traditional algorithms (PCA, LDA, LBP, SVM) to the deep learning models (CNN-based methods) under the influence of different real-world factors, is missing. The originality of this work is threefold: (i) the single taxonomy of face recognition development phases, (ii) the comparative performance analysis of the databases on a dataset basis using normalized measures of accuracy, and (iii) the detailed discussion of practical problems that include illumination variation, occlusion, age effects, and image resolution. Measured contributions can be made as a quantitative comparison across 4 benchmark datasets (AR, AT&T, Yale, FERET), showing that CNN-based models consistently achieve higher recognition accuracy (up to 99.79) than classical methods. The review presents researchers and practitioners with practical information on the choice of face recognition models that can be used in the real-world context.

Contrasting Evolutionary Algorithm and Social Spider Algorithm using Travelling Salesman Problem

Eman Arif, Asma Sanam Larik, Tasneem Adnan, Maryam Raees Ahmed — Fri, 31 Jul 2026 00:00:00 +0000

Combinatorial optimization problems have plenty of attention, both for the reasons of their complexity and for their practical applications. Looking at the range of metaheuristic techniques, there are two which have shown good results: the Genetic Algorithms (GA) and swarm-based techniques. A comparative study of the Genetic Algorithm and the Social Spider Algorithm (SSA) is described in this paper with Travelling Salesman Problem (TSP) as benchmark. The SSA, which is based on cooperative interaction mechanisms among the agents, is modified to address the TSP and its performance is compared to GA. The algorithms are contrasted with respect to the quality of solutions and efficiency of computation. The experimental results demonstrate that the convergence property and runtime efficiency of Social Spiders improves the problem considered. The paper is arranged as follows: In Section 1 and 2, is about introduction to the paperwork discussion of literature gaps section 3 we introduce the concept of Genetic algorithm and Swarm While Section 4, the problem is formulated. Section 5 focuses on implementation of Social Spider Algorithm and 6 presents experimental results and comparison of the two algorithms at hand: Genetic, Social Spider Optimization. Finally, in Section 6 conclusions are drawn up and prospects are discussed.

Towards Sustainable 5G Networks: Addressing Security, Environmental, and Deployment Challenges

Fri, 31 Jul 2026 00:00:00 +0000

The development from first-generation (1G) to fifth generation (5G) mobile networks has paved the way for improved connectivity, bandwidth, and latency. But the deployments of 5G have also posed many security concerns, architectural obstacles, and environmental sustainability issues. The current body of literature focuses on performance optimization or individual security solutions. Still, there is a need for a critical study of the lack of models that can simultaneously tackle the security, sustainability, and deployment challenges. This paper examines the most important potential threat vectors in the 5G environment, such as Software-Defined Networking (SDN) vulnerabilities and Network Function Virtualization (NFV) vulnerabilities, mobile cloud threats, multi-domain trust issues, IoT-based attack surfaces, and the environmental impact of large-scale infrastructure. We would like to propose a holistic solution to mitigate attacks based on SDN, NFV, Machine Learning (ML)-based intrusion detection, Mobile Cloud Computing (MCC), and blockchain-based decentralized trust management. A conceptual evaluation model is created to make comparisons between these approaches with respect to four dimensions – security effectiveness, scalability, latency sensitivity, and energy efficiency. The main features of the present invention are: (1) providing a systematic analytical comparison of the emerging 5G security mechanisms; (2) matching security objectives with the sustainable networking goals; and (3) a conceptual evaluation framework to validate multi-layered 5G protection strategies. Results show that the orchestration of intelligence in the deployment of secure and eco-friendly 5G networks hinges on the ability to manage trust in a decentralized manner, design energy-efficient networks, and integrate privacy-preserving distributed intelligence, including federated learning.

Smart AI Bus Travel Planner: An Intelligent Crowd Prediction and Comfort-Aware Recommendation System for Inter-City Public Transportation

Umm-e- Laila, Alisha Nasir, Saima Batool, Saifullah Khan, Syed Saad Ali — Fri, 31 Jul 2026 00:00:00 +0000

Inter-city buses are an important mode of transportation for millions of daily commuters, especially in
developing countries. However, existing transit systems provide limited information regarding passenger density, seat
availability, and overall travel comfort in real time. This lack of intelligent decision support often forces commuters
to make uncertain travel decisions, resulting in overcrowding, longer waiting times, and uncomfortable travel
experiences. This paper proposes an AI-Powered mobile application called Smart AI Bus Travel Planner that assists
commuters by predicting passenger, crowd level, calculating comfort score, recommending suitable seats, and
generating digital tickets automatically. Which uses machine learning techniques to improve the overall inter-city bus
travel experience. By predicting the number of people on the bus, scoring how comfortable the ride is, recommending
comfortable seats, and automatically generating a digital ticket. The system takes the raw travel logs, route data, and
simulated occupancy data, and feeds it into a custom-built four-module AI pipeline. The system uses a routine
detection algorithm to identify recurring travel patterns and a Random Forest regression model to predict crowd levels
categorized as Green, Yellow, or Red. In addition, based on the bus type and the number of passengers, a custom
formula is developed to calculate comfort scores, and a seat recommendation algorithm is designed that matches the
seat preferences of passengers with the gender-based seating constraints. The application is developed using a stack
of React Native for the front-end, Node.js for the back end, and MongoDB for the data. The app offers these insights
as well as real-time GPS tracking, QR code ticketing and interactive maps. The system was evaluated using 5,200
simulated trip records collected from 12 inter-city routes. The comfort score showed a strong correlation with user
satisfaction (R2= 0.84). And the user acceptance in a pilot study with 45 users is 91%. The proposed system introduces
predictive intelligence into inter-city transportation by providing crowd forecasting, comfort recommendations, and
automated travel planning features.

Analyze the Deep Learning Algorithms for Image Enhancement

Faiza Latif Abbasi, Dilbar Hussain, Saira Khurram Arbab — Fri, 31 Jul 2026 00:00:00 +0000

Image reconstruction and image enhancement are basic problems in computer vision, and are of great values for
medical image, satellite remote sensing and image restoration in digital photographs. The reconstruction quality is directly related
to the diagnostic reliability in medical imaging (CT, MRI, X-ray) and to environmental monitoring or in process decision making
in satellite remote sensing, in high stakes applications. Current traditional methods such as interpolation, iterative, and
optimization based methods have the disadvantage of being inherently limited with regards to computational efficiency, failure to
deal with complex profiles of noise and poor performance on low-resolution input data. This paper introduces an original deep
learning architecture that combines a modified 27-layer Convolutional Neural Network (CNN) with the DenseNet121 backbone in
a more synergistic way to overcome these shortcomings. The architecture proposed exploits the reuse of dense features and an
encoder-decoder structure to capture the local texture features and the global structure, hence, facilitating higher fidelity of the
image. The framework is conditioned and tested on the Deep Autoencoder Image Reconstruction benchmark dataset. The results of
the experiment show that the proposed method has an accuracy of 94.6%, a precision of 92.3%, a recall of 91%, and an F1-score
of 90.3%, which is better than the state-of-the-art conventional and deep learning baselines. The work contributions include the
architectural design, the CNN-integration strategy, and a highly rigorous evaluation protocol using multiple metrics. The
innovations have great potential in the further evolution of smart imaging systems in a wide range of real-world applications.