Publications

Peer-reviewed work in machine learning and applied AI.

IEEE2023

Design and Development of an Efficient Risk Prediction Model for Cervical Cancer

A Gradient Boosting model that predicts an individual’s risk of cervical cancer from lifestyle, clinical, and demographic factors (age, sexual history, HPV status, smoking). After preprocessing — handling missing values and building a combined cancer-status target from Dx:Cancer and Dx:CIN — features were ranked with XGBoost, and SMOTE oversampling plus stratified cross-validation addressed class imbalance, reaching 98.9% accuracy.

Read paper →
IEEE2024

DistilBERT-BiLSTM for Multi-Category Web Content Classification

A hybrid model pairing DistilBERT’s contextual embeddings with a bidirectional LSTM to sort web pages into up to ten categories (Science & Technology, NSFW, Entertainment, News, Fitness, and more). Combining transformer feature extraction with sequential modeling gives precise identification across text, audio, video, and image content, improving web filtering for all age groups.

Read paper →
MECS Press2025

Enhanced Credit Card Fraud Detection with Tuned Isolation Forest Ensemble

An ensemble of K-Nearest Neighbors, Random Forest, Logistic Regression, and an Isolation Forest, with the Isolation Forest’s hyperparameters tuned automatically via GridSearchCV, for real-time credit card fraud detection. Comparing the individual models against the tuned ensemble shows that combining classifiers and fine-tuning the Isolation Forest meaningfully improves accuracy and adaptability across different transaction datasets.

Read paper →