Gong Xudong — accepted AI-conference papers

2024

Goal-Conditioned On-Policy Reinforcement Learning

Gong Xudong, Feng Dawei, Kele Xu, Bo Ding, Huaimin Wang

NeurIPS 2024poster

Existing Goal-Conditioned Reinforcement Learning (GCRL) algorithms are built upon Hindsight Experience Replay (HER), which densifies rewards through hindsight replay and leverages historical goal-achieving information to construct a learning curriculum. However, when the task is characterized by a n…

Cited by 1 Source ↗PDF ↗Scholar ↗

2024

Iterative Regularized Policy Optimization with Imperfect Demonstrations

Gong Xudong, Feng Dawei, Kele Xu, Yuanzhao Zhai, Chengkang Yao, Weijia Wang +2

ICML 2024poster

Imitation learning heavily relies on the quality of provided demonstrations. In scenarios where demonstrations are imperfect and rare, a prevalent approach for refining policies is through online fine-tuning with reinforcement learning, in which a Kullback–Leibler (KL) regularization is often employ…

Cited by 4 Source ↗PDF ↗code ↗Scholar ↗