Unsupervised Object Interaction Learning with Counterfactual Dynamics Models
Authors: Jongwook Choi, Sungtae Lee, Xinyu Wang, Sungryull Sohn, Honglak Lee
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Through experiments on continuous control environments such as Magnetic Block and 2.5-D Stacking Box, we demonstrate that an agent can learn object interaction behaviors (e.g., attaching or stacking one block to another) without any external rewards or domainspeciļ¬c knowledge. |
| Researcher Affiliation | Collaboration | Jongwook Choi*1, Sungtae Lee*2, Xinyu Wang1, Sungryull Sohn3, Honglak Lee1,3 1University of Michigan 2Individual Researcher 3LG AI Research |
| Pseudocode | No | The paper describes algorithms and methods in prose and with mathematical formulations, but it does not include any explicitly labeled 'Pseudocode' or 'Algorithm' blocks or figures. |
| Open Source Code | No | The paper does not contain any explicit statement about releasing source code for the described methodology, nor does it provide a link to a code repository. |
| Open Datasets | No | The paper refers to custom-built 'continuous control environments' named 'Stacking Box' and 'Magnetic Blocks'. While these environments are described, there is no indication that they are publicly available as datasets, nor are any links, DOIs, or formal citations provided for public access to these environments/datasets. |
| Dataset Splits | No | The paper mentions 'evaluation episodes' and success rates, but it does not specify explicit training, validation, or test dataset splits (e.g., percentages, sample counts, or references to predefined splits) for reproducibility. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware used for running the experiments, such as GPU models, CPU types, or memory specifications. |
| Software Dependencies | No | The paper mentions software components like 'SAC (Haarnoja et al. 2018)' and 'scaled dot-product attention architecture (Vaswani et al. 2017)', but it does not provide specific version numbers for any programming languages, libraries, or solvers (e.g., Python, PyTorch, TensorFlow, CUDA versions). |
| Experiment Setup | No | The 'Implementation Details' section describes the network architecture (using attention) and mentions that SAC is used for the RL algorithm, but it does not provide concrete hyperparameter values (e.g., learning rate, batch size, number of epochs) or other system-level training settings. |