Izisekelo zamaNethiwekhi E-Neural Okwenziwayo
I-Artificial intelligence (AI) ikhona yonke indawo ezimpilweni zethu zansuku zonke, kusukela ekwazini izwi kuma-smartphones ethu kuya ekunconyweni kwemikhiqizo ezingosini zethu zokuthenga eziku-inthanethi. Kodwa lezi zinhlelo “zingafunda” futhi “zicabange” kanjani ngobunyoninco? Impendulo ilele kumanethiwekhi e-neural okwenziwa, futhi ikakhulukazi, Ekufundeni Okujulile.
Isifundo esithi “Qala Ekufundeni Okujulile” ku-OpenClassrooms sikungenisa emhlabeni othakazelisayo wamanethiwekhi emizwa. Uzothola i-neuron esemthethweni, isisekelo salawa manethiwekhi, nokuthi igqugquzelwa kanjani ukusebenza kwama-neuron ephilayo obuchopho bethu. Lawa ma-neurons ahlelekile, lapho exhunywe kunethiwekhi, angacubungula futhi ahumushe idatha eyinkimbinkimbi, kungaba izithombe, imisindo noma imibhalo.
Kodwa lokho kumane kuyisihloko se-iceberg. Uzophinda uhlole ukuthi lawa manethiwekhi akhiwe kanjani ngezandlalelo, okuvumela ukuhlaziya okujulile nokuyinkimbinkimbi kwedatha. Izakhiwo ezinjengama-autoencoder namanethiwekhi ajulile anezendlalelo eziguqukayo azokwambulwa, okukubonisa amandla nokuguquguquka kokufunda okujulile.
Ukufunda Okujulile: Inguquko Yezobunhloli Bokwenziwa
Ukufunda Okujulile, noma Ukufunda Okujulile, kungaphezu nje komkhuba wobuchwepheshe. Lokhu kuyimpumelelo enkulu eguqule umkhakha wobuhlakani bokwenziwa wabheka phansi. Kodwa yini eyenza i-Deep Learning ikhetheke kakhulu futhi ihluke kakhulu kwamanye amasu e-AI?
Okokuqala, i-Deep Learning iyakwazi ukucubungula amanani amakhulu edatha engahlelekile. Emhlabeni lapho ukuchofoza ngakunye, ukuthanda kwakho konke kukhiqiza idatha, leli khono libaluleke kakhulu. Amanethiwekhi e-neural ajulile, anezendlalelo zawo eziningi, angakwazi ukuhlonza amaphethini nezakhiwo kulawa madathasethi amakhulu lapho izindlela ezivamile zihluleka khona.
Khona-ke, Ukufunda Okujulile ukuzifundela wena. Kunokuba ifune ukuhlelwa okusobala, isebenzisa idatha ukuze iqeqeshe futhi ithuthuke. Lokhu kusho ukuthi uma idatha eningi evezwa kuyo, kuba ngcono ukusebenza kwayo. Lesi sici siyenza isebenze ngokukhethekile emisebenzini efana nokubonwa kwesithombe noma ukuhumusha ngomshini.
Ngaphezu kwalokho, Ukufunda Okujulile kuyisisekelo sezinto eziningi ezintsha zobuchwepheshe. Kusukela kubasizi bezwi abahlakaniphile kuya ezimotweni ezizishayelayo, yinjini eyenza lezi zintuthuko.
Ngokuhlola Ukufunda Okujulile, asitholi nje kuphela ubuchwepheshe, kodwa futhi nendlela entsha yokubona nokuqonda umhlaba. Inikeza umbono oyingqayizivele wokuthi imishini ingafunda kanjani, iguquke, futhi ekugcineni ihlanganyele nabantu ukuxazulula izinkinga eziyinkimbinkimbi.
Izicelo Ezisebenzayo Nekusasa Lokufunda Okujulile
Ukufunda Okujulile akugcini nje kuthiyori noma kumalabhorethri ocwaningo. Seluvele lugxile ezimpilweni zethu zansuku zonke, lunikeza izixazululo ezintsha ezinseleleni ebesicabanga ukuthi zingenakunqotshwa eminyakeni embalwa edlule.
Thatha isibonelo sezempilo. Amanethiwekhi ajulile e-neural asetshenziselwa ukuthola izifo ezithombeni zezokwelapha ngezinye izikhathi ngokunemba okungcono kunodokotela. Bangakwazi ukubona izimila, ukungajwayelekile kwenhliziyo, noma ezinye izimo ngokuhlaziya ama-X ray, ama-MRIs, noma izikena.
Emkhakheni wezobuciko, Ukufunda Okujulile kusetshenziselwa ukudala imisebenzi emisha noma ukubuyisela endala. Ama-algorithms angakha umculo, apende izithombe noma abhale izinkondlo, ecindezela imingcele phakathi komshini nokudala komuntu.
Imboni ayishiywa ngaphandle. Imigqa yokukhiqiza isebenzisa i-Deep Learning ukuze kuthuthukiswe izinqubo, ibone amaphutha noma ibikezele ukwehluleka kwezisetshenziswa. Lokhu kulungiselelwa kuholela ekusebenzeni kahle okukhulu, izindleko ezincishisiwe kanye nekhwalithi yomkhiqizo engcono.
Kodwa lokhu kumane kuyisiqalo. Ikusasa le-Deep Learning ligcwele isithembiso. Njengoba ama-algorithms eguquguquka futhi amandla ekhompuyutha enyuka, izinhlelo zokusebenza zesikhathi esizayo zingashintsha kakhulu. Kungakhathaliseki ukuthi ukuqonda izimfihlakalo zendawo yonke, ukulwa nokushintsha kwesimo sezulu noma ukudala amadolobha ahlakaniphile, i-Deep Learning izoba senhliziyweni yalezi ntuthuko, yakhe ikusasa elingcono labo bonke.
