1 paper · 1 filter
Mayank Daswani, Jan Leike
What is happiness for reinforcement learning agents? We seek a formal definition satisfying a list of desiderata. Our proposed definition of happiness is the temporal difference er…