Ịmụta ajụjụ
Ịmụ Q-Learning bụ nkwado mmụta algọridim na-akụziri onye ọrụ nke omume na-akwụ ụgwọ kacha mma site na iji nke nta nke nta mụta uru nke mmegharị ọ bụla site na nnwale na njehie.
Nchịkọta
It matters because it can find optimal behavior without ever being told the rules of its environment.
Ime miri emi
Q-Learning na-amụta ọrụ a na-akpọ Q(s, a): ụgwọ ọrụ ogologo oge a na-atụ anya ya nke ime ihe 'a' na steeti's' wee mee nke ọma ma emechaa. Onye nnọchi anya ahụ na-amalite ịma ihe ọ bụla, nwaa omume, wee hụ ụgwọ ọrụ. Mgbe usoro nke ọ bụla gasịrị, ọ na-atụgharị atụmatụ Q-uru ya gaa n'ụgwọ ọrụ enwetara gbakwunyere uru ebelatara kacha mma n'ọdịnihu ọ na-atụ anya na steeti na-esote. N'ụzọ dị mkpa, ọ bụ 'off-atumatu' na 'enweghị ihe nlereanya': ọ nwere ike ịmụta amụma kachasị mma mgbe ọ na-eme nchọpụta na-enweghị usoro, ọ dịghịkwa mkpa ihe nlereanya nke ka ụwa si agbanwe. Nyere nyocha zuru oke nke ụzọ omume steeti ọ bụla, ụkpụrụ Q na-agbakọ na ụkpụrụ kacha mma, yana ihe kacha mma na steeti ọ bụla bụ naanị nke nwere Q kacha elu.
Nghọta nka nka
Isi bụ mmelite Bellman: Q(s,a) <- Q(s,a) + alpha[r + gamma*max_a'Q(s',a') - Q(s,a)]. Alfa bụ ọnụ ọgụgụ mmụta, gamma ihe na-ebelata ego na-ebu ụgwọ ọrụ n'ọdịnihu, na okwu nke etinyere ya bụ njehie nwa oge-iche. 'Max' n'elu omume ọzọ bụ ihe na-eme ka ọ ghara ịdị n'usoro iwu ma mee ka ọ mụta amụma kacha mma anyaukwu ọbụna mgbe ọ na-eme nchọpụta. A na-ejikwa epsilon-anyaukwu eme ihe nyocha.
Mmetụta atụmatụ
Ọnụ ego na mmefu ego
Mkpebi ihe owuwu ụlọ na-akwalite arụmọrụ yana ọnụ ahịa ọrụ ruo ọtụtụ afọ.
Mkpebi doro anya
Nkà mmụta nka na-enyere ndị otu egwuregwu aka ịhọrọ nchịkọta ziri ezi, ọ bụghị naanị nke kachasị ọhụrụ.
Quality akara
Nhọrọ injinia ka mma na-ebelata ihe omume ntụkwasị obi na mmepụta.
Ọdịnihu nke mmụta Q
Q-Learning tabular Classic na-adọga mgbe steeti dị ọtụtụ ịchekwa na tebụl. Ntuziaka kachasị na-ejikọta ya na netwọkụ akwara ozi, dị ka ọ dị na Deep Q-Networks (DQN), nke dị ka ụkpụrụ Q si na ntinye raw dị ka pikselụ. Nnyocha na-aga n'ihu na-eme ka nke a guzosie ike site na ịmegharị ahụmahụ, netwọk lekwasịrị anya, na ụdị dị iche iche dị ka Double DQN na nkesa Q-Learning nke na-ebelata nhụsianya ngafe ma na-anọchi anya nkesa nloghachi zuru oke karịa otu nkezi.
Mmejuputa n'ezie n'ụwa
Ndị na-egwu egwuregwu Atari (DeepMind's DQN) na-amụta ịkpọ Breakout na Pong ozugbo site na pikselụ ihuenyo.
Na-ebuli oge ọkụ-ọkụ okporo ụzọ n'okporo ụzọ iji wedata mkpokọta oge nchere ụgbọ ala
Ntugharị robot site na grid ma ọ bụ maze ebe robot na-amụta ụzọ kacha nso na-ebuli ụgwọ ọrụ.
Mkpebi ọnụahịa dị egwu na ngwa ahịa ebe onye nnọchi anya na-amụta omume ndị na-ebuli uru na-adịte aka
Ihe ize ndụ & okporo ụzọ nche
Ịkwalite otu akara ngosi nwere ike zoo adịghị ike sistemụ sara mbara.
A na-eledakarị ihe akụrụngwa na ụgwọ ọrụ anya.
Ọdịiche nchekwa na nleba anya nwere ike itolite ka sistemu na-adịwanye mgbagwoju anya.
Map mmejuputa
Kọwaa latency, ịdịmma na ebumnuche ọnụ ahịa tupu mmejuputa ya.
Benchmark n'okpuru ibu dị adị na ọnọdụ data.
Nleba anya akụrụngwa maka mperi, ịkpafu na mmetụta onye ọrụ.
Kwadebe ụzọ nzaghachi azụghachi azụ na ihe omume tupu ịchachaa.
Nọgide na-eme nchọpụta
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 Q-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
Ntuziaka na-esote
Ịhazi ọnụ ọgụgụ mmụta
Ajụjụ a na-ajụkarị
What is Q-Learning?
Ịmụ Q-Learning bụ nkwado mmụta algọridim na-akụziri onye ọrụ nke omume na-akwụ ụgwọ kacha mma site na iji nke nta nke nta mụta uru nke mmegharị ọ bụla site na nnwale na njehie. Ọ dị mkpa n'ihi na ọ nwere ike ịchọta ezigbo omume na-agwaghị ya iwu nke gburugburu ebe obibi ya.
Gịnị ka Q-uru Q(s, a) na-anọchi anya?
Q(s, a) na-eme atụmatụ ngụkọta ụgwọ ọrụ mbelata n'ọdịnihu site n'ime ihe na steeti steeti yana ịkpa àgwà kachasị mma ma emesịa, ọ bụghị naanị ụgwọ ọrụ ozugbo.
Kedu ihe kpatara eji kọwaa Q-Learning dị ka 'ọpụpụ-atụmatụ'?
Nke kacha karịa omume ndị ọzọ pụtara Q-Learning na-amụta uru nke ezigbo anyaukwu nke amụma ọbụlagodi mgbe onye nnọchi anya na-eji usoro omume dị iche na-enyocha uru.
N'ime iwu nwelite, kedu ihe gamma mbelata ego na-achịkwa?
Gamma (n'etiti 0 na 1) na-ebelata ụgwọ ọrụ n'ọdịnihu; ụkpụrụ dị nso 1 na-eme ka onye ọrụ ahụ nwee anya nke ọma, ụkpụrụ dị nso 0 na-eme ka ọ ghara ịhụ ụzọ.
Kedu ihe bụ njehie nwa oge-iche (TD) na Q-Learning?
Njehie TD bụ ọdịiche dị n'etiti atụmatụ ebumnuche ọhụrụ (ụgwọ ọrụ gbakwunyere ọnụ ahịa mbelata kacha mma n'ọdịnihu) yana atụmatụ Q ochie; mmelite na-ebelata oghere a.
Kedu ihe kpatara Q-Learning tabular na-agba mgba na nnukwu nsogbu dị ka egwuregwu vidiyo sitere na pikselụ?
Tebụlụ nleba anya chọrọ ntinye otu ụzọ omume steeti ọ bụla, nke na-agaghị ekwe omume mgbe steeti ruru ijeri ijeri, na-akpali ndị na-eme ihe ntanetị dị ka DQN.