My research aims to analyze, understand, and improve deep learning models that are applied to or used by humans. My personal homepage is here.
publications:
- BERTnesia: Investigating the capture and forgetting of knowledge in BERT in the Proceedings of Workshop on BlackboxNLP (co-located in EMNLP 2020). Jonas Wallat, Jaspreet Singh, Avishek Anand.
Interests
- Probing deep learning models
- Interpretability for search and news recommendation
- AI for precision medicine