Publications
DistortBench: Benchmarking Vision Language Models on Image Distortion Identification
A benchmark for evaluating vision language models on their ability to identify and reason about image distortions.
arXiv →Global PIQA: Evaluating Physical Commonsense Reasoning Across 100+ Languages and Cultures
A multilingual and multicultural extension of the PIQA benchmark evaluating physical commonsense reasoning across 100+ languages.
arXiv →Curiosity-Driven LLM-as-a-Judge for Personalized Creative Judgment
A curiosity-based approach for using LLMs as judges to better align automated evaluation of creative content with human preferences.
arXiv →Automatic FAQ Generation Using Text-to-Text Transformer Model
End-to-end pipeline for automatic FAQ generation from text using T5-based text-to-text transformer models. Incorporates coherence metrics between sentences to improve coverage and relevance of generated QA pairs.
IEEE Xplore →Source Code Plagiarism Detection: A Machine Intelligence Approach
Deep learning techniques for detecting plagiarism in source code. Analyzed supervised and unsupervised ML algorithms and expanded to capture multisource plagiarism at a granular code level. Achieved accuracy in excess of 90%.
IEEE Xplore →Machine Learning Techniques for Multisource Plagiarism Detection
Extended plagiarism detection techniques to capture multisource plagiarism cases by analyzing source code at a granular level using supervised and unsupervised ML algorithms.
IEEE Xplore →