?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=Action+Selection+in+a+hypothetical+house+robot%3A+Using+those+RL+numbers&rft.creator=Humphrys%2C+Mark&rft.subject=Animal+Behavior&rft.subject=Ethology&rft.subject=Artificial+Intelligence&rft.subject=Dynamical+Systems&rft.subject=Machine+Learning&rft.subject=Robotics&rft.description=Reinforcement+Learning+(RL)+methods%2C+in+contrast+to+many+forms+of+machine+learning%2C+build+up+value+functions+for+actions.+That+is%2C+an+agent+not+only+knows+%60what'+it+wants+to+do%2C+it+also+knows+%60how+much'+it+wants+to+do+it.+Traditionally%2C+the+latter+are+used+to+produce+the+former+and+are+then+ignored%2C+since+the+agent+is+assumed+to+act+alone.+But+the+latter+numbers+contain+useful+information+-+they+tell+us+how+much+the+agent+will+suffer+if+its+action+is+not+executed+(perhaps+not+much).+They+tell+us+which+actions+the+agent+can+compromise+on+and+which+it+cannot.+It+is+clear+that+many+interesting+systems+possess+multiple+parallel+and+conflicting+goals%2C+all+demanding+attention%2C+and+none+of+which+can+be+fully+satisfied+expect+at+the+expense+of+others.+Animals+are+the+prime+example+of+such+systems.+In+%5BHumphrys%2C+1995%5D%2C+I+introduced+the+W-learning+algorithms%2C+showing+one+method+of+resolving+competition+among+behaviors+automatically+by+reference+to+their+RL+values.+The+scheme+has+the+unusal+feature+that+behaviors+are+at+all+times+in+selfish+pursuit+of+their+own+goals+and+have+no+explicit+concept+of+cooperation%2C+despite+residing+in+the+same+body.+In+this+paper%2C+I+apply+W-learning+to+the+world+of+a+hypothetical+house+robot%2C+which+doubles+as+family+toy%2C+movile+security+camera%2C+mobile+smoke+alarm+and+occasional+vacuum+cleaner.+I+show+how+a+W-learning+community+of+behaviors+inside+the+robot+will+support+a+robust+behavior+pattern%2C+capabable+of+opportunistic+behavior%2C+avoiding+dithering%2C+and+allowing+for+the+concept+of+default+behavior+and+expression+of+low-priority+goals.&rft.publisher=ICSC+Academic+Press&rft.contributor=Anderson%2C+Peter+G.&rft.contributor=Warwick%2C+Kevin&rft.date=1996&rft.type=Conference+Paper&rft.type=NonPeerReviewed&rft.format=application%2Fpostscript&rft.identifier=http%3A%2F%2Fcogprints.org%2F448%2F2%2Ff.SOCO96.ps&rft.identifier=++Humphrys%2C+Mark++(1996)+Action+Selection+in+a+hypothetical+house+robot%3A+Using+those+RL+numbers.++%5BConference+Paper%5D+++++&rft.relation=http%3A%2F%2Fcogprints.org%2F448%2F