《1995 A reinforcement Learning approach to job-shop scheduling》.pdf

A R e i n f o r c e m e n t L e a r n i n g A p p r o a c h t o J o b - s h o p S c h e d u l i n g We i Zhang T h o m a s G D i e t t e n c h Department of Compute r Science Department of Compute r Science Oregon Stat e Unjversit y Oregon Stat e Universit y Corvalhs , Orego n 97331-3202 Corvalhs , Orego n 97 511-3202 U S A U S A Abstrac t goal of tht learning syste m is to find a policy that max i mizes the expected reinforce ment over futur e actions In We apply reinforce merit learning methods to the context of jo b shop scheduling , (he policy tells what learn domain-specific heuristics fo r jo b shop scheduling actio n to make next i n order to maximiz e scheduling A repair-based scheduler starts some measure of the quality of the fina l schedule wit h a critical-pat h schedule and incrementally In this paper we focus o n the applicatio n domai n of repairs constraint violation s wit h the goal of space shuttle payload processing for NASA The. goal is finding a short conflict-free schedule The te m