Sie sind hier: Startseite Teaching SS2018 hrl-q5


Questions to HRL Paper 5

Q1: What is the problem with learning from sparse reward signals and how can auxiliary tasks facilitate the learning of the original task?

Q2: Informally describe the two parts of the proposed framework. Are the parameters of part one updated in the second part? Explain your answer.

Q3: What is the difference between SAC-U and SAC-Q and SAC-Q (pixels) and why does SAC-Q (pixels) take longer to learn?