PhD Student · IISc Bangalore

Abhipsa Basu

I work on fairness in deep learning models, geographical bias in text–to–image generation, and using diffusion-generated synthetic images to train fairer classifiers.

Email abhipsabasu@iisc.ac.in ★ Google Scholar </> GitHub PDF CV

01 / About

I am a PhD student in the Department of Computational & Data Sciences at the Indian Institute of Science, Bangalore, advised by Prof. R. Venkatesh Babu, and part of the Vision and AI Lab (VAL). Before this, I worked as a Cognitive Data Scientist at IBM India, and completed my M.Tech at IIT Kharagpur and B.Tech at Heritage Institute of Technology, Kolkata.

My research looks at where deep learning models pick up and amplify bias — from classifiers trained on group-imbalanced data, to text-to-image models that under-represent parts of the world — and how generative models themselves can be steered to produce fairer, more representative data.

Research interests

Fairness in deep learning models

Geographical knowledge of LLMs / VLMs

Enhancing fairness of synthetic images generated by diffusion models

Using synthetic images for fairness in downstream tasks

02 / Recent work

ICLR 2026