{"pk":31770,"title":"Thinking Locally to Act Globally: A Novel Approach to Reinforcement Learning","subtitle":null,"abstract":"Reinforcement Learning methods address the prob-\nlem faced by an agent w h o must choose actions in\nan u n k n o w n environment so as to maximize the re-\nwards it receives in return. T o date, the available\ntechniques have relied on temporal discounting, a\nproblematic practice of valuing immediate rewards\nmore heavily than future rewards, or else have im-\nposed strong restrictions on the environment. This\npaper sketches a n e w method which utilizes a subjec-\ntive evaluator of performance in order to (1) choose\nactions that maximize undiscounted rewards and (2)\ndo so at a computational advantage with respect to\nprevious discounted techniques. W e present initial\nexperimental results that attest to a substantial im-\nprovement in performance.","language":"eng","license":{"name":"","short_name":"","text":null,"url":""},"keywords":[],"section":"Submitted Presentations","is_remote":true,"remote_url":"https://escholarship.org/uc/item/91f167cn","frozenauthors":[{"first_name":"Anton","middle_name":"","last_name":"Schwartz","name_suffix":"","institution":"Stanford University","department":""}],"date_submitted":null,"date_accepted":null,"date_published":"1993-01-01T18:00:00Z","render_galley":null,"galleys":[{"label":"PDF","type":"pdf","path":"https://journalpub.escholarship.org/cognitivesciencesociety/article/31770/galley/22838/download/"}]}