Manuela Veloso

TL;DR: In this article , Monte-Carlo dropout is used to attain soft concept predictions that do not contain leaked information and aligning machine learning models with concepts that are understandable by human beings.

TL;DR: This work proposes and evaluates a novel Monte Carlo method for learning the scaling factors of subrewards, in which the training elicits humans' preferences between two state-action scenarios, and shows that the algorithm elicits preferences over explicit scenarios, it is less susceptible to human error than previous elicitation approaches.

TL;DR: Empirical results demonstrate that the Policy Library can be efficiently created and that the stored "eigen-policies" can be understood as a representation of the structure of the domain.

TL;DR: The report notes that the participants in the 2020 Science Group were geographically diverse, representing 12 nationalities, coming “from some of the world’s leading research institutions and companies [and]... elected for their expertise in a principal field”.