Novelty Search in Representational Space for Sample Efficient Exploration

Authors: Ruo Yu Tao, Vincent Francois-Lavet, Joelle Pineau

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test our approach on a number of maze tasks, as well as a control problem and show that our exploration approach is more sample-efficient compared to strong baselines.
Researcher Affiliation Academia Ruo Yu Tao1, 2, *, Vincent Franc ois-Lavet1, 2, Joelle Pineau1, 2 1 Mc Gill University 2 Mila, Quebec Artificial Intelligence Institute
Pseudocode Yes Algorithm 1: The Novelty Search algorithm in abstract representational space.
Open Source Code Yes The code with all experiments is available 1. 1https://github.com/taodav/nsrs
Open Datasets No The paper describes experiments on environments like 'Acrobot (Brockman et al., 2016)' and 'multi-step maze environment' which are either standard environments or custom-built, but it does not provide explicit access information (link, DOI, formal citation) to any publicly available *dataset* used for training.
Dataset Splits No The paper does not provide specific dataset split information (percentages, sample counts, or citations to predefined splits) for training, validation, or testing.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper mentions 'Pycolab game engine (Stepleton, 2017)' but does not provide specific version numbers for software dependencies or libraries used to replicate the experiment.
Experiment Setup Yes We discuss all environment-specific hyperparameters in Appendix J.