Koyon Ƙarfafawa Kan layi
Ƙarfafa ilmantarwa na kan layi yana horar da wakilai zalla daga ƙayyadaddun bayanai, da aka tattara a baya, ba tare da hulɗar kai tsaye da muhalli ba.
Dubawa
Yana da mahimmanci saboda a cikin kiwon lafiya, robotics, da shawarwari, bincike ta gwaji da kuskure yana da tsada, jinkirin, ko haɗari.
Zurfafa nutsewa
RL na kan layi (wanda kuma ake kira batch RL) yana koyon manufa daga madaidaicin tarihin abubuwan da suka gabata - jihohi, ayyuka, lada, da jihohi na gaba - ba tare da taɓa yin sabbin ayyuka a cikin yanayi na ainihi yayin horo ba. Wannan yana buɗe RL don saituna inda binciken kan layi ba shi da aminci ko tsada, kamar koyan manufofin jiyya daga bayanan haƙuri na tarihi ko ƙwarewar robot daga bayanan shiga. Matsalolin ƙayyadaddun ƙayyadaddun ƙayyadaddun ƙayyadaddun ƙayyadaddun ƙayyadaddun ƙayyadaddun ƙayyadaddun kuskure ne: daidaitattun hanyoyin da suka dogara da ƙima suna ƙima da ƙimar ayyukan da ba a rarrabawa ba wanda saitin bayanan bai taɓa gwadawa ba, kuma ba tare da yanayin gyara waɗannan kurakurai ba, manufar tana korar lada na yaudara. Algorithms na zamani suna fuskantar wannan ta kasancewa kusa da bayanan, ta amfani da ƙididdiga masu ra'ayin mazan jiya (CQL), ƙuntatawa na manufofin (BCQ, BEAR), ko ma'aunin nauyi (IQL).
Fahimtar Fasaha
Yanayin gazawar ainihin shine ƙima na ayyukan da ba a rarrabawa ba: aikin Q da aka koya yana ba da ƙima masu girma zuwa zaɓin aikin da ba ya nan daga saitin bayanai, kuma bootstrapping yana yada waɗannan kurakurai ba tare da ainihin ra'ayi don gyara su ba. Q-Learning Conservative (CQL) yana magance wannan ta ƙara mai ƙididdigewa wanda ke tura ƙimar Q-ƙimar ga ayyukan da ba a gani ba yayin da yake ci gaba da ɗaukar ayyukan cikin bayanai, yana samar da ƙananan iyaka akan ƙimar gaskiya da manufar da ke guje wa maras tallafi, zaɓin da ba a iya gani ba.
Dabarun Tasiri
Kudin da kasafin kuɗi
Hukunce-hukuncen gine-gine suna haifar da aiki da tsadar aiki na shekaru.
Shawarwari masu haske
Ilimin fasaha yana taimaka wa ƙungiyoyi su zaɓi tari mai kyau, ba kawai sabon abu ba.
Kula da inganci
Zaɓuɓɓukan injiniya mafi kyau suna rage abin dogaro a cikin samarwa.
Makomar Ƙarfafa Koyon Ƙarfafa Kan layi
RL na kan layi yana haɗuwa tare da ƙirar ƙira-hanyoyi kamar Mai Canjawa Mai Canjawa sake sake shi azaman tsinkayar ayyukan da aka tsara akan dawowar da ake so - kuma tare da babban horo, ba da damar jami'ai waɗanda aka horar da su akan manyan bayanan shigar da bayanai sannan zaɓin daidaitawa akan layi. Yi tsammanin ci gaba a cikin kiwon lafiya, tuƙi mai cin gashin kansa, da shawarwarin inda amintaccen koyo daga bayanan da ke akwai ke da mahimmanci, tare da ingantattun kayan aiki don kimanta manufofin layi don haka za a iya amincewa da manufofin da aka tura kafin su taɓa yin aiki a duniyar gaske.
Aiwatar da Gaskiyar Duniya
Koyon manufofin jiyya na asibiti daga bayanan lafiyar lantarki na tarihi
Horar da mutummutumi daga manyan bayanan da aka shigar ba tare da bincike mai haɗari ba
Haɓaka shawarwari da tsarin talla daga rajistan ayyukan hulɗar da suka gabata
Inganta manufofin yanke shawarar tuƙi mai cin gashin kai daga bayanan rundunar da aka tattara
Hatsari & Tsare-tsare
Haɓaka ma'auni ɗaya na iya ɓoye manyan raunin tsarin.
Sau da yawa ana raina kayan more rayuwa da kuma kuɗin kulawa.
Tsaro da gibin lura na iya girma yayin da tsarin ke ƙara haɓaka.
Taswirar Hanya
Ƙayyade latency, inganci, da maƙasudin farashi kafin aiwatarwa.
Alamar ma'auni a ƙarƙashin ainihin kaya da yanayin bayanai.
Kula da kayan aiki don kurakurai, ɗigo, da tasirin mai amfani.
Shirya bijirowa da hanyoyin mayar da martani kafin sikeli.
Ci gaba da Bincike
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Offline Reinforcement Learning quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Jagora na gaba
Ƙarfafa Koyo Daga Sauraron Dan Adam
Tambayoyin da ake yawan yi
Menene Ƙarfafa Ƙarfafawa Kan layi?
Ƙarfafa ilmantarwa na kan layi yana horar da wakilai zalla daga ƙayyadaddun bayanai, da aka tattara a baya, ba tare da hulɗar kai tsaye da muhalli ba. Yana da mahimmanci saboda a cikin kiwon lafiya, robotics, da shawarwari, bincike ta gwaji da kuskure yana da tsada, jinkirin, ko haɗari.
Menene ma'anar ƙarfafa ilmantarwa ta layi (batch)?
RL na kan layi yana koyon manufofin gaba ɗaya daga bayanan da aka tattara a baya kuma baya yin hulɗa da yanayi yayin horo.
Menene babban ƙalubalen fasaha a cikin RL na layi?
Hanyoyin ƙima suna ƙima da ayyukan da ba sa cikin bayanan, kuma ba tare da wani yanayi don gyara waɗannan kura-kurai ba manufar tana bin lada na yaudara.
Me yasa RL na layi yana da kyau don aikace-aikacen kiwon lafiya?
Binciken gwaji da kuskure akan marasa lafiya yana da haɗari, don haka koyan manufofin jiyya daga bayanan tarihi yana guje wa jefa mutane cikin haɗari.
Ta yaya Conservative Q-Learning (CQL) ke yaƙar ƙima?
CQL yana ƙara hukunci wanda ke rage ƙimar Q-ƙimar ayyukan da ba a cikin bayanan ba yayin da ake kiyaye ayyukan cikin-bayani babba, yana ba da ƙarancin ƙima akan ƙima.
Ta yaya Mai Canja wurin Yankewa yake gyara RL a layi?
Canjin yanke shawara yana ɗaukar matakai azaman jeri kuma yana haifar da sharuɗɗan ayyuka akan komawa-zuwa-tafi, sarrafa simintin gyare-gyare azaman tsinkayar jeri na autoregressive.