Decoupling Meta-Reinforcement Learning with Gaussian Task Contexts and Skills
Authors: Hongcai He, Anjie Zhu, Shuang Liang, Feiyu Chen, Jie Shao
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments in the navigation and robot manipulation continuous control tasks show that DCMRL is more effective than previous meta-RL methods with more generalizable prior experience. We evaluate DCMRL in two challenging continuous robot control environments, i.e., maze navigation and kitchen manipulation, which are long-horizon and sparse-reward. The results show that DCMRL outperforms previous meta-RL methods, achieving more effective adaptation to unseen target tasks. |
| Researcher Affiliation | Academia | 1University of Electronic Science and Technology of China, Chengdu, China 2Sichuan Artificial Intelligence Research Institute, Yibin, China {hehongc,anjiezhu}@std.uestc.edu.cn, {shuangliang,chenfeiyu,shaojie}@uestc.edu.cn |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. Figure 1 is a method overview diagram and Figure 3 is an architecture diagram. |
| Open Source Code | Yes | Our code is available at https://github.com/hehongc/DCMRL/. |
| Open Datasets | No | The paper mentions using a "task-agnostic dataset of state-action trajectories D" and evaluating in "maze navigation and kitchen manipulation" environments. While these environments are known, the paper does not provide concrete access information (link, DOI, specific repository, or formal citation for the dataset D itself) for a publicly available or open dataset. |
| Dataset Splits | No | The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning into train/validation/test sets. |
| Hardware Specification | No | The paper does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers like Python 3.8, PyTorch 1.x) needed to replicate the experiment. |
| Experiment Setup | No | The paper does not contain specific experimental setup details (concrete hyperparameter values, training configurations, or system-level settings) in the main text. It mentions that "More details about the experimental environments and baselines are in He et al. (2023)", referring to a different paper. |