Machine Theory of Mind
Authors: Neil Rabinowitz, Frank Perbet, Francis Song, Chiyuan Zhang, S. M. Ali Eslami, Matthew Botvinick
ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | This paper is structured as a sequence of experiments of increasing complexity on this Machine Theory of Mind network, which we call a To Mnet. These experiments showcase the idea of the To Mnet, exhibit its capabilities, and demonstrate its capacity to learn rich models of other agents incorporating canonical features of humans Theory of Mind, such as the recognition of false beliefs. |
| Researcher Affiliation | Industry | 1DeepMind 2Google Brain. Correspondence to: Neil Rabinowitz <ncr@google.com>. |
| Pseudocode | No | The paper describes the architecture and components of the To Mnet but does not include structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any explicit statements about releasing source code or links to a code repository for the methodology described. |
| Open Datasets | No | The paper describes generating environments ('gridworlds') and agents ('randomly sampling wall, object, and initial agent locations', 'trained using deep RL') but does not specify a publicly available dataset with concrete access information (link, DOI, or citation to an established public dataset). |
| Dataset Splits | No | The paper mentions holding out a test set but does not provide specific percentages, sample counts, or detailed methodology for training, validation, or test splits. For instance, it says 'Npast might vary, and may even be zero' for past observations, but not about data splits. |
| Hardware Specification | No | The paper does not specify any particular hardware (e.g., GPU, CPU models, or cloud instances) used for running the experiments. |
| Software Dependencies | No | The paper mentions 'All agents used the UNREAL architecture (Jaderberg et al., 2017, Appendix D)' but does not provide specific version numbers for any software dependencies or libraries. |
| Experiment Setup | No | Precise details of the architecture, loss, and hyperparameters for each experiment are given in Appendix A. While details are available in an appendix, the main text does not contain specific hyperparameter values or detailed training configurations. |