Zhenheng Yang

Papers

Total Citations

214

H-Index

About

Zhenheng Yang is a computer vision and machine learning researcher whose work sits at the intersection of temporal reasoning, video understanding, and predictive intelligence. He is best known for his pioneering contributions to **action anticipation** — the challenging task of detecting and predicting human actions *before* they occur, a capability with transformative implications for robotics, autonomous systems, and intelligent surveillance. His most influential work, **"RED: Reinforced Encoder-Decoder Networks for Action Anticipation"** (2017), garnered over 190 citations and introduced a novel deep learning framework that moves beyond passive action recognition. Rather than simply categorizing observed actions, RED leverages reinforcement learning within an encoder-decoder architecture to anticipate future visual representations and classify actions proactively. This approach addressed critical limitations in prior methods that relied on fully observed sequences, pushing the field toward truly predictive video analysis. Yang's research reflects a broader commitment to building AI systems that understand the *dynamics* of the world — not just what is happening, but what *will* happen. His work has meaningfully shaped how the computer vision community approaches temporal prediction, making him a notable figure for students and researchers working on video understanding, human-robot interaction, and real-time intelligent systems.