Publications
Peer-reviewed work in machine learning and applied AI.
Design and Development of an Efficient Risk Prediction Model for Cervical Cancer
A gradient boosting model that estimates someone’s risk of cervical cancer from lifestyle, clinical, and demographic factors like age, HPV status, and smoking. I cleaned the data, built a single cancer-status label from the Dx:Cancer and Dx:CIN columns, ranked features with XGBoost, and used SMOTE with stratified cross-validation to handle the class imbalance. It reached 98.9% accuracy.
Read paper →DistilBERT-BiLSTM for Multi-Category Web Content Classification
A model that runs DistilBERT embeddings through a bidirectional LSTM to sort web pages into up to ten categories (Science & Technology, NSFW, Entertainment, News, Fitness, and so on). Pairing the transformer with the sequence model handled text, audio, video, and image content well, which is useful for web filtering across age groups.
Read paper →Enhanced Credit Card Fraud Detection with Tuned Isolation Forest Ensemble
A fraud-detection setup that combines K-Nearest Neighbors, Random Forest, Logistic Regression, and an Isolation Forest, with the Isolation Forest tuned through GridSearchCV. Comparing each model on its own against the tuned ensemble showed the ensemble was more accurate and held up better across different transaction datasets.
Read paper →