Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Universal Value Iteration Networks: When Spatially-Invariant Is Not Universal
Authors: Li Zhang, Xin Li, Sen Chen, Hongyu Zang, Jie Huang, Mingzhong Wang6778-6785
AAAI 2020 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate UVIN with both spatially invariant and spatially variant tasks, including navigation in regular maze, chessboard maze, and Mars, and Minecraft item syntheses. Results show that UVIN can achieve similar performance as VIN and its extensions on spatially invariant tasks, and significantly outperforms other models on more general problems. |
| Researcher Affiliation | Academia | Li Zhang,1 Xin Li, 1 Sen Chen,1 Hongyu Zang,1 Jie Huang,1 Mingzhong Wang2 1School of Computer Science, Beijing Institute of Technology, China 2USC Business School, University of the Sunshine Coast, Australia |
| Pseudocode | Yes | Algorithm 1 UVIN training via reinforcement learning |
| Open Source Code | Yes | Please refer to https: //github.com/bit1029public/UVIN for the source codes and the videos of game playing with the policy computed. |
| Open Datasets | Yes | We validated UVIN in Mine RL (Guss et al. 2019) to play Minecraft games without extra training. |
| Dataset Splits | Yes | We used 80% of them as the training set, the remaining 20% as the test set. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., GPU model, CPU type) used for running the experiments. |
| Software Dependencies | No | The paper mentions 'Python 3.8' but does not list any specific software libraries, frameworks, or solvers with their version numbers required to reproduce the experiments. |
| Experiment Setup | Yes | In experiments, we set |K| = 9 as suggested in (Tamar et al. 2016). If the agent cannot arrive at the goal state within 250 steps, the episode also terminates. |