{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/uncertainty-aware-learning-from","title":"Uncertainty Aware Learning from Demonstrations in Multiple Contexts using Bayesian Neural Networks","arxiv_id":"1903.05697","date":"2019-03-13","proceeding":null,"authors":["Sanjay Thakur","Herke van Hoof","Juan Camilo Gamboa Higuera","Doina Precup","David Meger"],"abstract":"Diversity of environments is a key challenge that causes learned robotic\ncontrollers to fail due to the discrepancies between the training and\nevaluation conditions. Training from demonstrations in various conditions can\nmitigate---but not completely prevent---such failures. Learned controllers such\nas neural networks typically do not have a notion of uncertainty that allows to\ndiagnose an offset between training and testing conditions, and potentially\nintervene. In this work, we propose to use Bayesian Neural Networks, which have\nsuch a notion of uncertainty. We show that uncertainty can be leveraged to\nconsistently detect situations in high-dimensional simulated and real robotic\ndomains in which the performance of the learned controller would be sub-par.\nAlso, we show that such an uncertainty based solution allows making an informed\ndecision about when to invoke a fallback strategy. One fallback strategy is to\nrequest more data. We empirically show that providing data only when requested\nresults in increased data-efficiency.","url_abs":"http://arxiv.org/abs/1903.05697v1","url_pdf":"http://arxiv.org/pdf/1903.05697v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"uncertainty-aware-learning-from","repo_url":"https://github.com/sanjaythakur/Uncertainty-aware-Imitation-Learning-on-Multiple-Tasks-using-Bayesian-Neural-Networks","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"diversity","task_name":"Diversity"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1903.05697","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}