Recent News
Selected for Participation in ACM Winter School on EdgeAI 2025 IISC, Bangalore.
C3VLM: Compute-Efficient & Compressed Small VLMs for Commonsense VQA Proposal Selected For Poster Presentation at IndoML Graduate Forum 2025.
NLKI-Lightweight-Natural-Language-Knowledge-Integration-Framework Paper Accepted at EMNLP Findings 2025 Suzhou, China.
Delivered a talk and Hands on Text Analytics and Processing for the AICTE QIP PG Certificate course at IIT-Kharagpur.
Selected Publications
C3VLM: Compute-Efficient & Compressed Small VLMs for Commonsense VQA Proposal
Aritra Dutta
We propose two complementary directions to achieve compact, efficient, and dependable reasoning in Vision-Language Models (VLMs) under low-compute/low-latency constraints...
NLKI: A Lightweight Natural Language Knowledge Integration Framework for Improving Small VLMs in Commonsense VQA
Aritra Dutta, Swapnanil Mukherjee, Deepanway Ghosal, Somak Aditya
A Noise Robust Knowledge Integration Framework for Efficient Small Vision Language Models.
Teaching Assistance Experience
Programming and Data Structure Lab (CS19003)
Machine Learning (CS60050)
Experience & Education
MS in Computer Science
Advised by Prof. Somak Aditya. & Prof. Pawan Goyal Focused on compression and reasoning in Small Vision Language Models.
Junior Research Fellow
Researched multi-modal reasoning capabilities of Encoder only Architectures. Framed a new strategy for noise robustness in Visual Question Answering (VQA).
Asst. System Engineer
Contributed to the backend and performance testing of Health & Motor Insurance Portal.
B.Tech in Computer Science
GPA: 9.14/10. Undergraduate thesis on Object Detection and Recognition.