Hi there, I’m Ananya, a PhD student at the Department of Computer Science, IIT Madras working on problems in NLP and Machine Learning with the guidance of Dr Mitesh M. Khapra. My PhD thesis titled “Towards Better Evaluation of NLG systems” includes a survey of metrics used for various NLG tasks; their shortcomings such as susceptibility to adversarial attacks / adversarial scenarios and analysis of their formulation for caveats and biases; leveraging large-scale, task-specific pretraining for formulating more generalizable metrics for Dialogue systems that are robust to some of the synthetic perturbations; a perturbation checklist for evaluating various criteria / dimensions of multiple NLG tasks; and evaluation of multilingual models in languages beyond English (specifically Indian languages in the context of Machine Translation). I am grateful to be supported by Google PhD fellowship and Prime Minister’s Fellowship for Doctoral Research. I have also worked on Reading Comprehension / QA models, Code generation, Phonetic misspellings based on inter-language interferences during my internships.

My research interests include Generative AI and its Evaluation, Adversarial Attacks and Analysis, Interpretability, Human-in-the-loop techniques, Responsible AI (Fairness of models, Societal Implications, Informed usage in applications, etc), and AI for Social Good.

You can reach me at ananya[at]cse[dot]iitm[dot]ac[dot]in or locate me in RBCDSAI, IIT Madras.

News

• 20-02-2024 [Upcoming] Attending AAAI at Vancouver, Canada to present our paper ‘Closing the gap in the trade-off between fair representations and accuracy’ at Deployable AI workshop.
• 01-02-2024 Participated in Google research week at Banglore, India.
• 04-12-2023 Invited as Young Ambassador to particiapte in Global Technology Summit at New Delhi, India.
• 06-12-2023 Attended EMNLP 2023 at Singapore.