Kuyini Ukufunda Komshini?

Amakhomphyutha awawatholi kodwa ahlakaniphile nsuku zonke

Ngendlela elula, ukufundwa komshini (ML) kuyimishini yemishini (computer) ukuze isebenze umsebenzi oceliwe ngokusebenzisa nokuhlaziya idatha (ulwazi) ukwenza lo msebenzi ngokuzimela, ngaphandle kokufaka okungeziwe okuvela kumthuthukisi womuntu.

Ukufunda komshini 101

Igama elithi "ukufundwa komshini" lahlanganiswa kuma-IBM labakwa-IBR ngo-1959 ngu-Arthur Samuel, iphayona elisebenza ngobuhlakani (AI) nokudlala kwekhompyutha. Ukufunda komshini, ngenxa yalokho, kuyigatsha le-Intelligence Intelligence. Isisekelo sikaSamuel kwakuwukuthi flip imodeli ye-computing yesikhathi esibheke phansi bese uyeka ukunikeza amakhompyutha izinto okufanele zifunde.

Esikhundleni salokho, wayefuna amakhompiyutha aqale ukuzicabangela izinto ngokwawo, ngaphandle kokuba abantu bafaka ngisho nocezu oluncane kakhulu lwolwazi. Khona-ke, wacabanga ukuthi amakhompiyutha ngeke asebenze kuphela kodwa angaqeda ukuthi yimiphi imisebenzi yokwenza nokuthi nini. Kungani? Ngakho amakhompyutha anganciphisa inani lomsebenzi abantu abadinga ukuwenza kunoma iyiphi indawo enikeziwe.

Indlela Ukufunda Kokusebenza Kanjani

Ukufunda komshini kusebenza ngokusetshenziswa kwe-algorithms nedatha. I-algorithm iyisethi yemiyalelo noma iziqondiso ezitshela ikhompyutha noma uhlelo ukuthi ungayenza kanjani umsebenzi. Ama-algorithms asetshenziswe ku-ML aqoqa idatha, aqaphe amaphethini, futhi asebenzise ukuhlaziywa kwalowo datha ukuvumelanisa nezinhlelo zawo nemisebenzi ukuqedela imisebenzi.

I-algorithms ye-ML isebenzisa izintambo zokubusa, izihlahla zesinqumo, amamodeli wezobuciko, ukucubungula kolimi lwemvelo, namanethiwekhi amanethiwekhi we-neural (ukubiza abambalwa) ukuzenzekelayo ukucubungula idatha ukwenza izinqumo nokwenza imisebenzi. Ngenkathi i-ML ingaba isihloko esiyinkimbinkimbi, Umshini Wokufundiswa we-Google inikeza izandla ezenziwe lula ngokubonisa ukuthi i-ML isebenza kanjani.

Indlela enamandla kakhulu yokufunda imishini esetshenziswa namuhla, ebizwa ngokuthi ukufunda okujulile , yakha isakhiwo esiyinkimbinkimbi semathempeli esibizwa ngokuthi inethiwekhi ye-neural, esekelwe emininingwaneni eningi yedatha. Amanethiwekhi e-Neural amaqoqo okulungiswa kwe-ML ne-AI ahlotshaniswa ngemuva kwamangqamuzana omjovo ebuchosheni bomuntu nolwazi lwenqubo yesistimu.

Intelligence Intuthuko vs. Ukufunda Machine vs. Data Mining

Ukuze uqonde kangcono ubuhlobo obuphakathi kwe-AI, i-ML, nokumbiwa kwedatha, kuyasiza ukucabanga ngesethi sezambulela ezihlukahlukene. I-A yisona isambulela esikhulu kunazo zonke. Isambulela se-ML isayizi encane futhi ifanelana ngaphansi kwesambulela se-AI. Isambulela semayini ye-mining iyona encane futhi ifanelana ngaphansi kwesambulela se-ML.

Imuphi Umshini Wokufunda Ongakwenza (futhi Usenayo kakade)

Umthamo wamakhompiyutha ukuhlaziya ulwazi oluningi emaqenjini wesibili kwenza i-ML isebenziseke ezimbonini ezimbalwa lapho isikhathi nokunemba kubalulekile.

Cishe usuvele uhlangabezane neML kaningi ngaphandle kokuqaphela. Ezinye zezinto ezijwayelekile zokusetshenziswa kobuchwepheshe be-ML zihlanganisa ukuqashelwa kwenkulumo okusebenzayo (i- Samsung's Bixby , i-Apple Siri , kanye nezinhlelo eziningi zokukhuluma-kuya manje ezijwayelekile kuma-PC), ukuhlunga ugaxekile kwe-imeyili yakho, ukwakha izindaba zezindaba, ukuthola ukukhwabanisa, ukuzenzela izincomo zezitolo, nokuhlinzeka ngemiphumela yokusesha ye-web esebenzayo.

I-ML ibandakanyeka nokudla kwakho kwe- Facebook . Uma uthanda noma uchofoze kokuthunyelwe komngane njalo, izilungiswa kanye ne-ML ngemuva kwezithombe "funda" ezenzweni zakho ngokuhamba kwesikhathi ukubeka phambili abanye abangane noma amakhasi ku-Newsfeed yakho.

Imuphi Umshini Wokufunda Ungakwazi & # 39;

Noma kunjalo, kunemingcele kulokho okungenziwa yi-ML. Isibonelo, ukusetshenziswa kobuchwepheshe be-ML ezimbonini ezahlukene kudinga inani elibalulekile lokuthuthukiswa kanye nokuhlelwa yizinhlelo zabantu ngokukhethekile uhlelo noma uhlelo lwezinhlobo zemisebenzi edingekayo yile mboni. Ngokwesibonelo, esibonelweni sethu sezokwelapha ngenhla, isimiso se-ML esasetshenziselwa umnyango ophuthumayo sakhiwe ngokuqondile ngemithi yabantu. Okwamanje akunakwenzeka ukuthatha lolu hlelo ngqo futhi luqalise ngokuqondile esikhungweni esiphuthumayo sezilwane. Ukuguquka okunjalo kudinga ukuhlaziywa okubanzi nokuthuthukiswa kwabasebenzi bezinhlelo zabantu ukudala inguqulo ekwazi ukwenza lo msebenzi ngemithi yezilwane noma yezilwane.

Kudinga futhi idatha enkulu kakhulu kanye nezibonelo 'zokufunda' ulwazi oludinga ukwenza izinqumo nokwenza imisebenzi. Izinhlelo ze-ML nazo zikhona ngempela ekuchazeni idatha futhi zilwa nomfanekiso kanye nezinye izinhlobo zobudlelwane phakathi kwemiphumela yedatha, njengokubangela nemiphumela.

Ukuthuthukiswa okuqhubekayo, noma kunjalo, okwenza ubuchwepheshe be-ML bube ngaphezu kokudala amakhompyutha anamandla nsuku zonke.