Papers
1
Total Citations
7
H-Index
1
About
Jianan Han is a researcher whose work lies at the intersection of video prediction, representation learning, and disentanglement—critical areas for advancing how machines understand and anticipate dynamic visual scenes. In their most-cited paper, "Taylor saves for later: Disentanglement for video prediction using Taylor representation" (2021, 7 citations), Han introduces a novel framework that leverages Taylor series expansions to disentangle spatial and temporal features in video data. This approach enables more accurate and interpretable future frame predictions by separating static content from evolving motion, a fundamental challenge in computer vision. The work demonstrates how mathematical tools can be repurposed for deep learning, offering a principled way to model complex temporal dependencies. While still early in their career, Han’s contribution stands out for its conceptual elegance and potential to influence subsequent research in video understanding and generative models. Their focus on disentangled representations aligns with broader efforts to build more robust and explainable AI systems, making Han a promising voice in the field.
Research Focus
Key Achievements
Top Papers
- 1