Publications

* denotes equal contribution

VIB-AVSR: Variational Information Bottleneck for Noise-Robust LLM-Based Audio-Visual Speech Recognition

P. Arora*, N. Singh*, U. Cappellazzo, S. Petridis, M. Pantic

INTERSPEECH 2026 (Oral) Paper

BLADE: Bias-Linked Adaptive DEbiasing

P. Arora*, N. Singh*, V. Diwan*, P. Mazumder

Under review Paper

When Big Models Train Small Ones: Label-Free Model Parity Alignment for Efficient Visual Question Answering with Small VLMs

A. Subramanyam*, N. Singh*, P. Arora*, A. Mishra

EMNLP 2025 Paper Code

Hybrid Sample Synthesis-Based Debiasing of Classifier in Limited Data Setting

P. Arora*, P. Mazumder*

WACV 2024 Paper arXiv Code