EN-HUI YANG

Full Professor@University of Waterloo·加拿大·OpenReview

研究方向

Information theory · multimedia compression · digital communications · source and channel coding · image and video coding · deep learning · information security

5.5

Differentiable JPEG-based Input Perturbation for Knowledge Distillation Amplification via Conditional Mutual Information Maximization

ICLR 2026Poster

通讯

5.0

Soft Quantization Activation Functions For Deep Learning

ICLR 2026Withdrawn

一作

6.8

Coupled Data and Measurement Space Dynamics for Enhanced Diffusion Posterior Sampling

NeurIPS 2025Poster

三作

6.5

Going Beyond Feature Similarity: Effective Dataset distillation based on Class-aware Conditional Mutual Information

ICLR 2025Poster

通讯

4.0

Generalization Error Minimized Deep Learning

合作者 (14)

Shayan Mohajer Hamidi

EN-HUI YANG

Differentiable JPEG-based Input Perturbation for Knowledge Distillation Amplification via Conditional Mutual Information Maximization

Soft Quantization Activation Functions For Deep Learning

Coupled Data and Measurement Space Dynamics for Enhanced Diffusion Posterior Sampling

JPEG Inspired Deep Learning

Enhancing Adversarial Robustness Through Robust Information Quantities

Towards Undistillable Models by Minimizing Conditional Mutual Information

Going Beyond Feature Similarity: Effective Dataset distillation based on Class-aware Conditional Mutual Information

Generalization Error Minimized Deep Learning