Policy Gradient Coagent Networks. Thomas, P. S abstract bibtex We present a novel class of actor-critic algorithms for actors consisting of sets of interacting modules. We present, analyze theoretically, and empirically evaluate an update rule for each module, which requires only local information: the module’s input, output, and the TD error broadcast by a critic. Such updates are necessary when computation of compatible features becomes prohibitively difficult and are also desirable to increase the biological plausibility of reinforcement learning methods.
@article{thomas_policy_nodate,
title = {Policy {Gradient} {Coagent} {Networks}},
abstract = {We present a novel class of actor-critic algorithms for actors consisting of sets of interacting modules. We present, analyze theoretically, and empirically evaluate an update rule for each module, which requires only local information: the module’s input, output, and the TD error broadcast by a critic. Such updates are necessary when computation of compatible features becomes prohibitively difficult and are also desirable to increase the biological plausibility of reinforcement learning methods.},
language = {en},
author = {Thomas, Philip S},
pages = {9}
}
Downloads: 0
{"_id":"FNcasneZcCJMxpFD9","bibbaseid":"thomas-policygradientcoagentnetworks","authorIDs":[],"author_short":["Thomas, P. S"],"bibdata":{"bibtype":"article","type":"article","title":"Policy Gradient Coagent Networks","abstract":"We present a novel class of actor-critic algorithms for actors consisting of sets of interacting modules. We present, analyze theoretically, and empirically evaluate an update rule for each module, which requires only local information: the module’s input, output, and the TD error broadcast by a critic. Such updates are necessary when computation of compatible features becomes prohibitively difficult and are also desirable to increase the biological plausibility of reinforcement learning methods.","language":"en","author":[{"propositions":[],"lastnames":["Thomas"],"firstnames":["Philip","S"],"suffixes":[]}],"pages":"9","bibtex":"@article{thomas_policy_nodate,\n\ttitle = {Policy {Gradient} {Coagent} {Networks}},\n\tabstract = {We present a novel class of actor-critic algorithms for actors consisting of sets of interacting modules. We present, analyze theoretically, and empirically evaluate an update rule for each module, which requires only local information: the module’s input, output, and the TD error broadcast by a critic. Such updates are necessary when computation of compatible features becomes prohibitively difficult and are also desirable to increase the biological plausibility of reinforcement learning methods.},\n\tlanguage = {en},\n\tauthor = {Thomas, Philip S},\n\tpages = {9}\n}\n\n","author_short":["Thomas, P. S"],"key":"thomas_policy_nodate","id":"thomas_policy_nodate","bibbaseid":"thomas-policygradientcoagentnetworks","role":"author","urls":{},"downloads":0,"html":""},"bibtype":"article","biburl":"https://bibbase.org/zotero/asneha213","creationDate":"2019-09-27T03:40:25.854Z","downloads":0,"keywords":[],"search_terms":["policy","gradient","coagent","networks","thomas"],"title":"Policy Gradient Coagent Networks","year":null,"dataSources":["fjacg9txEnNSDwee6"]}