Publications

Action Inference by Maximising Evidence: Zero-Shot Imitation from Observation with World Models

Published in NeurIPS 2023, 2023

We propose Action Inference by Maximising Evidence (AIME), a novel algorithm for imitation from observation by transferring learnt world models.

Recommended citation: Xingyuan Zhang, Philip Becker-Ehmck, Patrick van der Smagt, Maximilian Karl, Action Inference by Maximising Evidence: Zero-Shot Imitation from Observation with World Models, NeurIPS 2023 https://openreview.net/forum?id=WjlCQxpuxU

NeoRL: A near real-world benchmark for offline reinforcement learning

Published in NeurIPS 2022, Datasets and Benchmarks Track, 2022

NeoRL presents conservative datasets for offline RL, highlights the complete pipeline for deploying offline RL in real-world applications, and also benchmarks recent offline RL algorithms on NeoRL under the complete pipeline.

Recommended citation: Qin, R. J., Zhang, X., Gao, S., Chen, X. H., Li, Z., Zhang, W., & Yu, Y. (2022). NeoRL: A near real-world benchmark for offline reinforcement learning. Advances in Neural Information Processing Systems, 35, 24753-24765. https://openreview.net/pdf?id=jNdLszxdtra

Differentiable Spatial Regression: A Novel Method for 3D Hand Pose Estimation

Published in IEEE Transactions on Multimedia, 2020

We propose a novel framework for hand pose estimation from depth image. The framework mainly utilise a differentiable decoder structure.

Recommended citation: X. Zhang and F. Zhang, "Differentiable Spatial Regression: A Novel Method for 3D Hand Pose Estimation," in IEEE Transactions on Multimedia, vol. 24, pp. 166-176, 2022, doi: 10.1109/TMM.2020.3047552. https://ieeexplore.ieee.org/document/9309323