Bo-Han Wu
Papers
2
Total Citations
82
H-Index
2
About
Bo-Han Wu is a researcher advancing the frontiers of video prediction and generative modeling, with a focus on enabling intelligent agents to plan and act in complex, real-world environments. His most impactful work, "Greedy Hierarchical Variational Autoencoders for Large-Scale Video Prediction" (2021), has garnered over 75 citations, addressing a critical bottleneck in AI: the severe underfitting of video prediction models when applied to diverse, large-scale scenes. Wu’s key contribution lies in developing a hierarchical variational autoencoder framework that greedily scales to handle high-dimensional video data, significantly improving generalization beyond small, controlled datasets. This innovation is pivotal for robotics and autonomous systems, where accurate long-term video prediction is essential for tasks like navigation and manipulation. Beyond this flagship paper, his research continues to explore efficient, scalable architectures for spatiotemporal modeling. Wu’s work stands out for its practical impact, bridging the gap between theoretical generative models and real-world deployment, and has been recognized as a foundational step toward robust, planning-capable AI agents.
Research Focus
Key Achievements
Top Papers
- 1
- 2