About Me

School of Computer Science

Major: Evolutionary Computation and Intelligent Systems (ECIS) Lab

GPA: 3.88 / 4.30

Project: Development of a Web-Based Project Management Platform using Waterfall Technique

GPA: 3.49 / 4.00

Developed a knowledge distillation framework using neural network and random forest teachers to generate KDE-targeted synthetic data, improving symbolic regression performance under limited and unevenly distributed training data. Showed that KDE-based region identification outperforms geometric bounding box methods for extrapolation, with statistically significant improvements across six benchmark datasets. Conducted the first systematic comparison of five model selection criteria (MSE, AIC, BIC, MDL, PSM) for symbolic regression across 20 PMLB benchmark datasets, using Spearman rank correlation to measure alignment with test performance. Developed hybrid model selection criteria that extend standard metrics with extrapolation divergence and input sensitivity penalties; identified two practically useful configurations (MVP-BIC and MVP-MDL; MVP: Multiple Variable Perturbation) that consistently improve extrapolation model selection. Validated the hybrid framework on a real-world energy system surrogate modelling problem, achieving approximately 98% improvement in test MSE without domain-specific tuning. Achieved peer-reviewed publications in top venues including EuroGP, GECCO, and ECTA. Teaching Assistant for undergraduate and postgraduate modules for two years; did a weekly journal club among PhD researchers in the same field throughout the PhD. PhD Thesis: Symbolic Regression with Genetic Programming: Synthetic Data Generation and Model Selection for Interpolation and Extrapolation