Tsetlin Machine for Solving Contextual Bandit Problems

Authors: Raihan Seraj, Jivitesh Sharma, Ole-Christoffer Granmo

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our empirical analysis shows that Tsetlin Machine as a base contextual bandit learner outperforms other popular comparable base learners on eight out of nine datasets.
Researcher Affiliation Academia Raihan Seraj Electrical and Computer Engineering Mc Gill University Canada raihan.seraj@mail.mcgill.ca Jivitesh Sharma Center for Artificial Intelligence Research (CAIR) University of Agder Norway jivitesh.sharma@uia.no Ole-Christoffer Granmo Center for Artificial Intelligence Research (CAIR) University of Agder Norway ole.granmo@uia.no
Pseudocode Yes Algorithm 1 Complete WTM learning process; Algorithm 2 Thompson Sampling with TM
Open Source Code Yes The code is available online on: github
Open Datasets Yes We use standard supervised learning classification datasets from UCI machine learning repository [10]: Iris, Breast Cancer Wisconsin (Diagnostic), Adult, Covertype, and Statlog (Shuttle). ... Additionally we considered two other classification datasets: MNIST [9] and Noisy XOR... Movielens 100 K [15] and Simulated Article [23].
Dataset Splits No The paper states that training details including data splits are provided in Appendix B, but Appendix B is not included in the provided text. The main body of the paper does not explicitly detail the specific training, validation, or test data splits used for the experiments.
Hardware Specification No The paper explicitly states in its self-assessment checklist (3d) that it did not include the total amount of compute and the type of resources used for the experiments.
Software Dependencies No The paper does not provide a reproducible description of ancillary software, specifically lacking version numbers for key software components or libraries used for the experiments.
Experiment Setup Yes Details of binarization and parameter choices are provided in Appendix B.