Ankur Biswas

@mca.uem.edu.in

Assistant Professor, Department of Computer Applications
University of Engineering and Management, Kolkata

Ankur Biswas


                  Download Compact PDF  
https://researchid.co/ankur_biswas

RESEARCH, TEACHING, or OTHER INTERESTS

Computer Science, Artificial Intelligence, Computer Science Applications, Information Systems
8

Scopus Publications

Scopus Publications

  • Prediction of Groundwater Depth Using Average Rainfall-Groundwater Correlations in Kolkata, India
    Ankur Biswas, Palash Kundu, Piya Khan, Suryadipta Bhattacharyya, Sharmi Khan
    Lecture Notes in Networks and Systems, 2025
  • Visualizing Human Brain Networks for Mental Health Diagnosis
    Rittika Paul, Bristi Sengupta, Ankur Biswas, Dimple Mundhra, Nandini Samdariya, Anirban Das, Suvam Manna
    2025 International Conference on Engineering Innovations and Technologies Icoeit 2025, 2025
    The visualization of human brain networks using neuroimaging techniques such as electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) plays a crucial role in advancing mental health diagnosis. This paper introduces a hybrid AI framework that combines graph neural networks for connectivity analysis and Fourier-based spectral features to improve diagnostic precision. By applying graph theory and small-world network analysis, we decode structural and functional connectivity patterns and achieve high classification accuracy in identifying disorders such as schizophrenia and depression through multi-modal EEG-fMRI fusion. Unlike traditional unimodal methods, our approach integrates explainable AI to provide interpretable results for clinicians. We also explore challenges related to dataset standardization and machine learning integration. Future directions involve real-time edge computing for wearable devices and addressing ethical concerns in AI-driven clinical diagnostics.
  • Spam Protector: A Machine Learning-Based Multi-model Email Spam Protection System with Gmail IMAP Support
    Khushi Dugar, Swagata Sen, Viswajeet Sarangi, Roshani Raushan, Santanu Mahala, Ankur Biswas, Anirban Das
    2025 International Conference on Engineering Innovations and Technologies Icoeit 2025, 2025
    This study introduces a Gmail-integrated spam detection system that utilizes machine learning, accessed through the IMAP protocol. The primary goal is to deliver a user-friendly web application capable of effectively identifying and filtering spam emails. A hard voting ensemble classifier is proposed, combining the predictive capabilities of four distinct machine learning models: Naive Bayes, Support Vector Machine (SVM), Random Forest, and Logistic Regression. Each classifier is individually trained and assessed, with results benchmarked against the ensemble model's performance. The hard voting strategy yields a notable accuracy of approximately ninety six percent, surpassing the results of each model when used independently. The system is trained on a dataset of labeled emails drawn from a standard spam classification corpus. Evaluation metrics include accuracy, precision, recall, and f1-score. The findings illustrate the potential of ensemble learning in improving spam detection and provide a scalable, practical tool for users seeking reliable email filtering.
  • Optimizing Pharmaceutical Affordability: A Composition-Based Algorithm for Medicine Price and Alternative Discovery
    Jeet Karmakar, Shreeyoshi Goldar, Sourabh Maity, Komal Agarwal, Subhajit Kundu, Ankur Biswas, Anirban Das
    2025 International Conference on Engineering Innovations and Technologies Icoeit 2025, 2025
    There is still a big challenge in the availability of cheap medicines, especially in a country like India, which has a huge amount of out-of-pocket health expenditure. Patients have difficulty in choosing an affordable substitute for a proprietary medicine, as there exists very little or no credible information regarding such medicines. This paper advocates for an evidence-based recommendation system for low-priced medicinal substitutes, taking into consideration active ingredients, cost, credibility of the drug manufacturer, and packaging. With an annotated dataset of over 2.5 Lakh drugs from the Indian market, the system utilizes a lightweight algorithm to recommend and rank these alternatives. These suggestions maintain therapeutic equivalence without compromising on drug efficacy, thereby offering economical and lawful treatment options. The system is designed to offer real-time recommendations with very low overhead in terms of computational resources, making it most appropriate for rural clinics, community pharmacies, and mobile applications. In the evaluation stage, it was pitted against deep learning techniques: it provided competitive performance with greater speed and ease of use. Case studies with drugs such as Paracetamol and Ibuprofen denote the possibility of saving above 60%. The invention merges algorithmic accuracy to meet specific public health requirements, thus increasing fulfillment with treatment, reducing expenses, and enabling fair health care. Future updates can see machine learning, coupled with user input, to improve personalization and accuracy.
  • Revolutionizing Astrophysics: Black Holes, Gravitational Waves, and the Future of Cosmic Exploration
    Soumen Samanta, Bijen Sardar, Subhasree Paul, Rudradeep Debnath, Shrestha Gupta, Sagar Kumar Dhawa, Ankur Biswas, Anirban Das
    2025 International Conference on Engineering Innovations and Technologies Icoeit 2025, 2025
    This paper explores the revolutionary intersection of black hole physics and gravitational wave astronomy, representing one of the most dynamic frontiers in modern astrophysics. We review how gravitational wave observations provide unprecedented opportunities to explore the nature of black hole event horizons, potentially resolving long-standing theoretical paradoxes surrounding singularities where general relativity breaks down. The paper examines how gravitational wave “ringdown” signatures following black hole mergers may distinguish between competing theoretical models of black hole interiors, including predictions from string theory and loop quantum gravity. We analyze the distinctive gravitational wave patterns that might signal the existence of alternative compact objects predicted by modified gravity theories. Furthermore, we discuss how next-generation gravitational wave detectors will enable increasingly precise tests of the black hole paradigm, potentially revealing quantum gravity effects at black hole horizons. This research demonstrates how gravitational wave astronomy has transitioned from a purely theoretical concept to a powerful tool for observation. It introduces a new type of cosmic messenger, one that goes beyond electromagnetic radiation. This advancement may help resolve some of the most profound mysteries in modern physics.
  • Adaptive GAN-Based Watermarking for Robust Deepfake Detection
    Saradra Chanda, Manojit Das, Sudipta Maity, Bipul Naskar, Aritra Ghosh, Ankur Biswas, Tuhin Subhra Panda, Anirban Das
    2025 International Conference on Engineering Innovations and Technologies Icoeit 2025, 2025
    Progressions in Generative Adversarial Networks (GANs) have amplified the threat of deepfakes, challenging digital trust and authenticity. This paper proposes a novel GAN-based visible watermarking framework that uses a generator-discriminator layer in a looping fashion to embed indistinguishable, anti-tamper patterns into video frames. Spatial and temporal features are extracted through CNN and RNN architectures, while attention mechanisms specify the manipulated facial regions. Evaluated on FaceForensics++, Celeb-DF, and DFDC datasets using six frameworks. Unlike existing approaches, it effectively detects both watermarked and nonwatermarked deepfakes and resists compression and motionrelated tampering. This work presents a scalable, interpretable detection pipeline combining proactive watermarking with passive feature-based analysis.
  • MATRA: An Automated System for MATernal Risk Assessment
    Amartya Chakraborty, Stobak Dutta, Ankur Biswas, Prasenjit Das, Surendra Nath Bhagat, Subhankar Guha
    Smart Innovation Systems and Technologies, 2023
  • On Exploring the Role of Feature Processing in Gait-based Gender Identification
    Amartya Chakraborty, Stobak Dutta, Surendra Nath Bhagat, Subhankar Guha, Ankur Biswas, Parnava Roy
    Proceedings 2021 19th Oits International Conference on Information Technology Ocit 2021, 2021
    The advent of Internet of Things (IoT) in all domains of human life has made life smarter and simpler. The augmentation of machine intelligence in solving research problems focused on decision-making, has enriched the outcome of such research. In the field of biometric identification, IoT based research has seen rapid progress. Apart from the conventional biometric features, such as iris, finger-tips, etc., the possibility of using gait features from human walking activity has also been explored as a potential biometric feature. In the proposed work, a standard dataset gathered using heterogeneous sensors from Male and Female volunteers has been used. The experiments explore how the feature-scaling and Principal Component Analysis (PCA) methods enable the development of a fool-proof automated system based on Multi-Layer Perceptron (MLP), for identifying Male and Female gender based on gait signature.