Koyon da aka samo asali
 Koyon da aka samo asali | |
|---|---|
learning approach (en)  | |
| Bayanai | |
| Fuskar | machine learning (en) |
A cikin ilmantarwa na inji, ilmantarwa ta misali (wani lokacin ana kiranta ilmantarwa mai tushe ) iyali ne na algorithms na ilmantarwa wanda, maimakon yin bayyane, kwatanta sabbin lokuta na matsala tare da lokuta da aka gani a horo, waɗanda aka adana a cikin ƙwaƙwalwar ajiya. Saboda ana jinkirta lissafi har sai an lura da sabon misali, ana kiran waɗannan algorithms a wasu lokuta "lazy".
An kiran misali-tushen sabO yana gina ra'ayoyi kai tsaye daga lokuta na horo da kansu. Wannan yana nufin cewa rikitarwa na ra'ayi na iya girma tare da bayanai: [1] a mafi munin yanayin, ra'ayi shine jerin n abubuwan horo kuma rikitarwa ta lissafi na rarraba sabon misali shine O (n). Ɗaya daga cikin fa'idodi da ilmantarwa na misali ke da shi a kan wasu hanyoyin ilmantarwa ta inji shine ikonta na daidaita samfurinsa zuwa bayanan da ba a gani ba a baya. Masu koyo na asali na iya adana sabon misali ko jefa tsohon misali.
Misalan algorithms na ilmantarwa na misali sune algorithm na maƙwabta mafi kusa, injunan kernel da Cibiyoyin sadarwa na RBF. Wadannan adanawa (wani ɓangare na) tsarin horar da su; lokacin da suke hango darajar / aji don sabon misali, suna lissafin nesa ko kamanceceniya tsakanin wannan misali da lokuta na horo don yanke shawara.
Don yaƙi da rikitarwa na ƙwaƙwalwar ajiya na adana duk lokuta na horo, da kuma haɗarin wuce gona da iri ga hayaniya a cikin saitin horo, an ba da shawarar algorithms na rage misali.[2]
Dubi kuma
[gyara sashe | gyara masomin]- Misali na analog