@xuanalogue Move away from ANNs altogether. There are powerful approaches to program induction that learn programs directly in a programming language (mainly functional and logic programming languages). But, for those who prefer baby steps, here's some poi
116 followers
158 followers
RT @gajcinj: How can we verify that RL agents are making decisions for the right reasons and not relying on spurious correlations? Our pape…
708 followers
RT @gajcinj: How can we verify that RL agents are making decisions for the right reasons and not relying on spurious correlations? Our pape…
44 followers
How can we verify that RL agents are making decisions for the right reasons and not relying on spurious correlations? Our paper (me and @ivanadusparic) ''ReCCoVER: Detecting Causal Confusion for Explainable RL'' is out now in EXTRAAMAS post-proceedings. h