This is not an argument per se against using reinforcement learning, I am simply pointing out that you do in fact need to explicitly consider what your algorithm converges to in the limit rather…

Posters: jd_pressman. jd_pressman

3 likes recorded by the source corpus.