Mohammad Emtiyaz Khan
Papers
1
Total Citations
7
H-Index
1
About
Mohammad Emtiyaz Khan is a leading researcher in machine learning, whose work bridges Bayesian inference, optimization, and deep reinforcement learning. He is best known for pioneering **functional regularization** techniques that stabilize and accelerate training in deep Q-learning, directly addressing the limitations of traditional target networks. His 2021 paper on this topic, which has garnered 7 citations, proposes a novel method that improves reward propagation without sacrificing stability—a key contribution to modern reinforcement learning. Beyond this, Khan has made foundational contributions to **variational inference** and **continual learning**, developing algorithms that enable models to learn efficiently from streaming data without catastrophic forgetting. His research often emphasizes **Bayesian deep learning** and **natural-gradient methods**, offering principled frameworks for uncertainty estimation and optimization. With a citation count exceeding 7,000, his work has profoundly influenced both theory and practice, earning him recognition as a thought leader in probabilistic machine learning. Khan’s ability to distill complex mathematical ideas into practical, scalable algorithms makes his research essential reading for students and practitioners alike.
Research Focus
Key Achievements
Top Papers
- 1