- Umshini we-Teachable uvumela ukuthuthukiswa kwamamodeli okuqaphela isithombe, umsindo, kanye nokuma ngaphandle kokubhala ikhodi.
- Leli thuluzi livumela ubuhlakani bokwenziwa ukuba bube lula ngokwenza ukufunda komshini kube lula kumaphrofayili angewona obuchwepheshe kanye nabafundisi.
- Kungenzeka ukuthumela ngaphandle amamodeli aqeqeshiwe ukuze ahlanganiswe nezinhlelo zokusebenza zewebhu, amaphrojekthi e-IoT, noma isofthiwe yendawo.
Wake wacabanga ukuthi ubuhlakani bokwenziwa bungobochwepheshe bekhompyutha noma izazi zezibalo ezineziqu ze-PhD kuphela? Kulungile, susa lowo mbono ekhanda lakho ngoba namuhla noma ubani angazama ukufunda komshini ngaphandle kokulwa nomugqa owodwa wekhodi. Ngenxa yamathuluzi anembile, ukufinyelela kulobu buchwepheshe kuye kwaba yintando yeningi, okuvumela wonke umuntu kusukela kubafundi kuya kosomabhizinisi ukuthi bathathe izinyathelo zabo zokuqala ezweni le-AI ngendlela ebonakalayo ngokuphelele.
Esinye sezibonelo eziqhakazile kule ndawo yi-Teachable Machine, ipulatifomu yewebhu ye-Google eguqula ukuqeqeshwa kwamamodeli kube yinto efana nomdlalo. Akukhona ngokudala i-chatbot eyinkimbinkimbi, kodwa ngokuklama izinhlelo ezikwazi ukuhlukanisa izithombe, ukubona imisindo, noma ukubona ukuma komzimba ngesikhathi sangempela. Kuyindawo efanele yokungena kulabo abafuna ukuqonda ukuthi umshini "ufunda" kanjani ezibonelweni nokuthi bangalusebenzisa kanjani lolo lwazi kumaphrojekthi emfundo noma okudala.
Imiqondo eyisisekelo yokugwema ukulahleka endleleni
Ngaphambi kokungena ngaphakathi kwethuluzi, kuyasiza ukuqonda ukuthi kuyini i-Machine Learning . Empeleni, igatsha le-AI elivumela amakhompyutha ukuthi afunde kudatha esikhundleni sokulandela imiyalelo ehlelwe kusengaphambili isinyathelo ngesinyathelo. Cabanga ukuthi ungumntwana ofunda ukuhlukanisa i-apula nepheya; abazali bakho bakukhombisa izithelo ezahlukahlukene futhi bakutshele ukuthi yikuphi. Nokho, imodeli ye-AI yenza into efanayo: ihlaziya amaphethini kudatha enikeziwe ukuze yenze izinqumo ngezinto engakaze izibone ngaphambili.
Lawa mamodeli angamandla ashukumisayo obuchwepheshe esibusebenzisa nsuku zonke, njengabasizi bezwi bohlobo lwe-Alexa noma izinhlelo zokuncoma ze-Netflix. Emkhakheni wezemfundo, lokhu kubaluleke kakhulu, njengoba kuvumela abafundi ukuqonda ukusebenza kwangaphakathi kwe-AI nokuthuthukisa amakhono okucabanga ngokujulile mayelana nendlela ulwazi olucutshungulwa ngayo esikhathini sedijithali.
Isinyathelo ngesinyathelo ukudala imodeli yakho yokuqala
Uma ufuna ukuqala, inqubo ilula ngokumangalisayo. Okokuqala, iya kuwebhusayithi esemthethweni ye-Teachable Machine bese ukhetha uhlobo lwephrojekthi olukuthakazelisa kakhulu: izithombe, imisindo, noma ukuma . Uma ukhetha izithombe, isibonelo, uzobona ukuthi uhlelo lukucela ukuthi uchaze "amakilasi." Ikilasi limane nje liyisigaba ofuna i-AI isiqaphele. Uma ufuna ukuthi ihlukanise phakathi kwepeni nencwadi yamanothi, uzodala ikilasi ngalinye.
Ukuze wondle lezi zifundo, unezinketho ezimbili: sebenzisa ikhamera yakho yewebhu ukuze uthwebule izithombe ngesikhathi sangempela noma ulayishe amafayela osuvele uwagcinile kukhompyutha yakho noma ku-Google Drayivu. Icebo lokuqinisekisa ukuthi imodeli inembile ukunikeza ukuhlukahluka; ungathathi izithombe eziyishumi ezifanayo, kodwa kunalokho thwebula into kusuka kuma-engeli ahlukene, ukukhanya, kanye nezikhundla . Uma umshini unezibonelo ezihlanzekile nezihlukahlukene, kulapho uzowenza khona amaphutha ambalwa.
Uma idatha yakho isilungile, sekuyisikhathi sokuchofoza inkinobho ethi "Lungisa imodeli" . Yilapho umlingo uvela khona: uhlelo lucubungula izithombe bese luqala ukuqeqeshwa. Kubalulekile ukungavali ithebhu noma ukuvula izinhlelo ezidinga izinsiza eziningi phakathi nale nqubo ukuze ugweme amaphutha. Ngemuva kwalokho, ungahlola umphumela ekubukeni kuqala ngokubeka into phambi kwekhamera bese ubona ukuthi i-AI inika kanjani ithuba lokuphumelela esigabeni ngasinye.
Amathiphu athuthukisiwe kanye nokwenza ngcono idatha
Akusebenzi kahle ngaso sonke isikhathi okokuqala, futhi lokho kuyingxenye yobumnandi. Uma ubona imodeli yakho idideka, cishe udinga ukuhlanza isethi yakho yedatha . Isibonelo, uma uhlukanisa izinambuzane futhi kukhona umuntu ngemuva kwezithombe zezintuthwane, i-AI ingase icabange ukuthi abantu bayingxenye yalokho okuchaza izintuthwane. Ukususa lezi zithombe ezingabalulekile kubalulekile ekutholeni idatha "ehlanzekile" nokuqeqeshwa okusebenzayo.
Kulabo abafuna indlela yobungcweti, amathuluzi afana nesandiso se-Imageye Chrome akuvumela ukuthi ulande izithombe ngobuningi kusuka ku-Google ukuze wondle amamodeli. Ngaphezu kwalokho, kunconywa kakhulu ukubhukha amasampula athile ukuze ahlolwe ; okungukuthi, gcina izithombe ezimbalwa i-AI engazibonanga ngesikhathi sokuqeqeshwa ukuqinisekisa ukuthi ifunde ngempela ukwenza izithombe zibe zijwayelekile noma imane izibambe ngekhanda.
Izinhlelo zokusebenza zokuthumela ngaphandle nezomhlaba wangempela
Ingxenye engcono kakhulu ukuthi umsebenzi wakho akudingeki ugcinwe kwiwebhusayithi ye-Google. Ungathumela imodeli yakho ngefomethi ye-TensorFlow.js , okukuvumela ukuthi uyihlanganise kuwebhusayithi yakho noma kuhlelo lokusebenza lweselula. Kukhona nezinketho zokuyilanda ngamafomethi ahambisana ne-TensorFlow Lite, efanelekile kumaphrojekthi e-Internet of Things (IoT) noma kumadivayisi endawo.
Ngokombono wokufundisa, othisha bangasebenzisa lokhu ukudala imisebenzi ehlobene nezimo zangempela. Isibonelo, bangaqeqesha i-AI ukuhlonza izinhlobo ze-galaxy ngokusekelwe ku-morphology kaHubble, noma badale uhlelo oluqaphela izinto zansuku zonke ukuxazulula inkinga ethile yekilasi. Lokhu kuguqula abafundi ekubeni ngabathengi abangenzi lutho bobuchwepheshe babe ngabadali abasebenzayo bezixazululo ezisekelwe ku-AI.
Amanye amathuluzi abantu abangebona ochwepheshe
Nakuba i-Teachable Machine imnandi kakhulu, kunezinketho eziningi emakethe kulabo abafuna indlela yokusebenzisa ikhodi ephansi . I-Google AutoML iyisihlobo esikhulu, esinamandla kakhulu, futhi esigxile ebhizinisini, esihlangana kalula ne-Google Cloud. Ngakolunye uhlangothi, i-Lobe.ai inikeza ulwazi olulula kakhulu lokuhudula nokulahla lwamamodeli ezithombe, kuyilapho i-Runway ML iyindawo yokufihla abaculi bedijithali abafuna ukudala ubuciko noma ukuhlela amavidiyo nge-AI.
Uma ukhululekile kakhulu nge-Microsoft, i-Azure Machine Learning Studio ikuvumela ukuthi uklame imisebenzi eyinkimbinkimbi ngaphandle kokubhala ikhodi. Futhi uma ulimi nombhalo kuyinto oyithandayo, i-Hugging Face iyipulatifomu ehamba phambili, enikeza umtapo omkhulu wamamodeli aqeqeshwe kusengaphambili ongawahlela ngokwezifiso ukuze uhlaziye imizwa noma uhumushe izilimi. Zonke lezi zinkundla zinomgomo ofanayo: ukususa izithiyo zobuchwepheshe ukuze ukusungula izinto ezintsha kuncike emibonweni, hhayi ekwazini ukuthi ungabhala kanjani ikhodi.
Ikhono lokuqeqesha izinhlelo ezihlakaniphile selifinyeleleke kuwo wonke umuntu, okuvumela ubuhlakani kanye nemfundo ukuthi kuhlangane nobuchwepheshe obusezingeni eliphezulu. Ngokwazi amathuluzi afana ne-Teachable Machines kanye nezinye izindlela zawo, noma yimuphi umsebenzisi angaguqula idatha elula ibe amamodeli asebenzayo akwazi ukubona indawo akuyo, athuthukise izinqubo, futhi avule umnyango wezinto ezintsha ezazigcinelwe ochwepheshe bamakhompyutha kuphela.