I-VISual AI GUIDE

Amamodeli Okukhiqiza Asuselwa kumaphuzu

Amamodeli akhiqizayo asuselwa kumaphuzu adala idatha ngokufunda i-gradient yokusabalalisa idatha - isiqondiso esenza noma iyiphi isampuli enomsindo ibukeke kakhulu njengedatha yangempela.

Uhlolojikelele

Amamodeli akhiqizayo asuselwa kumaphuzu adala idatha ngokufunda i-gradient yokusabalalisa idatha - isiqondiso esenza noma iyiphi isampuli enomsindo ibukeke kakhulu njengedatha yangempela. Lokhu kubuka komsebenzi wamaphuzu kuhlanganisa amamodeli okusabalalisa ngezibalo ezihlukene ze-stochastic futhi kusekela amajeneretha ezithombe eziningi zesimanje.

Amamodeli Okukhiqiza Asuselwe Emaphuzuni awokugeleza komsebenzi okubonwa ngekhompuyutha atolika noma akhiqize imidiya ebonakalayo ukuze ihlaziywe, isebenze, futhi isungule.

I-Deep Dive

Esikhundleni sokumodela ngokuqondile, amamodeli asekelwe kumaphuzu afunda isikolo: igradient yokuminyana kwamathuba elogi ngokuphathelene nokokufaka. Ukwazi ukuthi iyiphi indlela yokugudluza isampula ukuze kwandiswe amathuba akhona kwanele ukukhiqiza idatha entsha. Umsebenzi ka-Yang Song kanye noStefano Ermon wango-2019 uqeqeshe inethiwekhi ukuze ilinganisele lesi sikolo kumaleveli amaningi omsindo kusetshenziswa ukufanisa amaphuzu we-denoising, kwase kwenziwa amasampula anamandla we-Langevin - enyathela amaphuzu ngokuphindaphindiwe futhi engeza umsindo omncane. Iphepha labo lango-2021 lamaphuzu-SDE libonise ukuthi amamodeli asuselwa kumaphuzu ahlukene anobuso obubili benqubo efanayo eqhubekayo echazwe isibalo sokuhlukanisa esitokisini. Okubalulekile ukuthi, yonke i-SDE ine-ODE 'yokugeleza kwamathuba' ahambisanayo abelana ngamamajini afanayo, evumela amathuba okuba khona kanye namasampula asheshayo.

I-Technical Insight

Ukulinganisa amaphuzu edatha ehlanzekile ngokuqondile kunzima lapho idatha iyingcosana, ngakho-ke imodeli iqeqeshwa kudatha ephazanyiswa umsindo we-Gaussian ezikalini eziningi. Ukufanisa amaphuzu we-denoising kunikeza ithagethi ethathekayo: isikolo sokusatshalaliswa komsindo silingana nesiqondisindlela somsindo esihlukaniswa nokuhluka komsindo, ngakho ukubikezela umsindo nokubikezela amaphuzu empeleni kuyinto efanayo. Ukusampula kuxazulula i-SDE yesikhathi esibuyela emuva (noma i-ODE yamathuba alinganayo) kusukela kumsindo we-Gaussian omsulwa.

I-Mastering Score-based Generative Models

Ukuze wakhe ukuqonda okujulile, phatha amamodeli akhiqizayo asekelwe ku-Score njengemodeli yokusebenza, hhayi isici esisodwa. Chaza imiphumela oyifunayo, cacisa ukucabanga, futhi uhlukanise lokho isistimu engakwenza ngokwethembeka kulokho okusadinga ukwahlulela kochwepheshe.

Empeleni, amaqembu aqinile asebenzisa ukunemba kwebhalansi yebhalansi yamamodeli asekelwe ku-Score-Based Generative namaqiniso okusebenza njengekhwalithi yedatha, ukuhluka kokukhanya, nokuvumelana kwamalebula. Babhala imibandela yempumelelo ecacile, ukuhlola okuqhathaniswa nedatha engokoqobo nokugeleza komsebenzi, futhi baphindaphinde ngokusekelwe kumaphethini okuhluleka aqashiwe esikhundleni sokuwina kwebhentshimakhi yesikhathi esisodwa. Yilapho ukuqonda kwethiyori kuguquka kube amandla ahlala njalo kuwo wonke umkhiqizo, inqubomgomo, kanye nokusebenza.

I-Visual AI ingakwazi ukuhlola, ukutholwa, nokumaka imisebenzi esikalini. Ngesikhathi esifanayo, amalungelo ezithombe kanye nemvume kungaba ubungozi bomthetho uma ukutholakala kungacacile. Indlela eqine kakhulu iwukuhlanganisa isivinini sokuhlola nesiyalo sokuphatha: qhuba abashayeli bezindiza, bamba ubufakazi, ushicilele amalogi ezinqumo, futhi ubuyekeze izivikelo ngokuqhubekayo njengoba imodeli yokuziphatha, okulindelwe ngabasebenzisi, kanye nezimfuneko zokulawula zishintsha.

I-Strategic Impact

I-Visual AI ingakwazi ukuhlola, ukutholwa, nokumaka imisebenzi esikalini.

I-Visual AI ingakwazi ukuhlola, ukutholwa, nokumaka imisebenzi esikalini. Ekusetshenzisweni kwekhwalithi ephezulu, lokhu kuhunyushwa emithethweni yokusebenza elinganisekayo, imingcele yobunikazi, nemikhuba yokubuyekeza ephindelelayo ukuze amaqembu akwazi ukukala ukuzethemba esikhundleni sokukala ukungaqondakali.

Amathimba aqanjiwe angakwazi ukulinganisa imiqondo ngokushesha ngezibuyekezo ezimbalwa ezenziwa mathupha.

Amathimba aqanjiwe angakwazi ukulinganisa imiqondo ngokushesha ngezibuyekezo ezimbalwa ezenziwa mathupha. Ekusetshenzisweni kwekhwalithi ephezulu, lokhu kuhunyushwa emithethweni yokusebenza elinganisekayo, imingcele yobunikazi, nemikhuba yokubuyekeza ephindelelayo ukuze amaqembu akwazi ukukala ukuzethemba esikhundleni sokukala ukungaqondakali.

Imisebenzi ingasebenzisa amasiginali wesithombe nawevidiyo obekunzima ukuwenza ngaphambilini.

Imisebenzi ingasebenzisa amasiginali wesithombe nawevidiyo obekunzima ukuwenza ngaphambilini. Ekusetshenzisweni kwekhwalithi ephezulu, lokhu kuhunyushwa emithethweni yokusebenza elinganisekayo, imingcele yobunikazi, nemikhuba yokubuyekeza ephindelelayo ukuze amaqembu akwazi ukukala ukuzethemba esikhundleni sokukala ukungaqondakali.

Ikusasa Lamamodeli Okukhiqiza Asuselwe Kwisikolo

Uhlaka lwamaphuzu-SDE luyinjini yethiyori engemuva kwenqubekelaphambili enkulu ye-AI ekhiqizayo. Izixazululi zezinombolo ezisheshayo, amashejuli omsindo angcono, kanye ne-ODE yokugeleza kwamathuba kunika amandla isizukulwane sesikhathi sangempela kanye nokuhlola ukuthi kungenzeka yini. Umbono ofanayo wokumatanisa amaphuzu usabaleleka ngale kwezithombe uye ekwakhiweni komsindo, i-molecular and protein structure, point cloud, kanye nokulingiswa kwesayensi, kuyilapho amamodeli afanayo kanye nokugeleza okuhambisanayo akhela ngokuqondile kulezi zisekelo zesikhathi esiqhubekayo ukuze inciphe ukukhiqiza ukuya ezinyathelweni ezimbalwa.

Ukuqaliswa Komhlaba Wangempela

I-Noise-Conditional Score Networks (NCSN) ekhiqiza ubuso obubonakalayo ngokulandela ama-gradients afundiwe nge-Langevin dynamics.

Ukwakhiwa kabusha kwesithombe sezokwelapha, njenge-MRI esheshisiwe, lapho isikolo esifundiwe sisebenza njengangaphambi kokugcwalisa idatha yokuskena engasampulanga kangako.

Ukukhiqizwa kwesakhiwo samangqamuzana namaprotheni ekutholweni kwezidakamizwa, ukumodela ukulungiselelwa kwe-athomu ye-3D ngokusatshalaliswa okusekelwe kumaphuzu.

Ukwakheka kwe-wave wave yomsindo lapho amamodeli wamaphuzu ezwakala ebhekise enkulumweni ehlanzekile noma emculweni, njengamavokhoda asekelwe ekusakazweni.

Amaphethini Okusebenzisa

Amamodeli Okukhiqiza Asuselwa kumaphuzu asebenza

I-Noise-Conditional Score Networks (NCSN) ekhiqiza ubuso obubonakalayo ngokulandela ama-gradients afundiwe nge-Langevin dynamics.

Amaqembu ngokuvamile athola imiphumela engcono uma echaza izilinganiso zekhwalithi ngaphambili, agcina indlela yokukhuphuka yomuntu yamakesi asemaphethelweni, futhi alandelele kokubili izinzuzo zokukhiqiza nezindleko zamaphutha ngokuhamba kwesikhathi.

Amamodeli Okukhiqiza Asuselwa kumaphuzu asebenza

Ukwakhiwa kabusha kwesithombe sezokwelapha, njenge-MRI esheshisiwe, lapho isikolo esifundiwe sisebenza njengangaphambi kokugcwalisa idatha yokuskena engasampulanga kangako.

Amaqembu ngokuvamile athola imiphumela engcono uma echaza izilinganiso zekhwalithi ngaphambili, agcina indlela yokukhuphuka yomuntu yamakesi asemaphethelweni, futhi alandelele kokubili izinzuzo zokukhiqiza nezindleko zamaphutha ngokuhamba kwesikhathi.

Amamodeli Okukhiqiza Asuselwa kumaphuzu asebenza

Ukukhiqizwa kwesakhiwo samangqamuzana namaprotheni ekutholweni kwezidakamizwa, ukumodela ukulungiselelwa kwe-athomu ye-3D ngokusatshalaliswa okusekelwe kumaphuzu.

Amaqembu ngokuvamile athola imiphumela engcono uma echaza izilinganiso zekhwalithi ngaphambili, agcina indlela yokukhuphuka yomuntu yamakesi asemaphethelweni, futhi alandelele kokubili izinzuzo zokukhiqiza nezindleko zamaphutha ngokuhamba kwesikhathi.

Amamodeli Okukhiqiza Asuselwa kumaphuzu asebenza

Ukwakheka kwe-wave wave yomsindo lapho amamodeli wamaphuzu ezwakala ebhekise enkulumweni ehlanzekile noma emculweni, njengamavokhoda asekelwe ekusakazweni.

Amaqembu ngokuvamile athola imiphumela engcono uma echaza izilinganiso zekhwalithi ngaphambili, agcina indlela yokukhuphuka yomuntu yamakesi asemaphethelweni, futhi alandelele kokubili izinzuzo zokukhiqiza nezindleko zamaphutha ngokuhamba kwesikhathi.

Izingozi & Guardrails

!

Amalungelo ezithombe kanye nemvume kungaba ubungozi bezomthetho uma ukuvela kungacacile.

!

Ukusebenza kwemodeli kungahluka kukho konke ukukhanya, izibalo zabantu, kanye nezindawo.

!

Okuhle okungelona iqiniso kungase kungabonakali ngaphandle uma izinga lokuzethemba liqashelwa.

Ukuqalisa Umhlahlandlela

1

Chaza indlela yokwamukela yokunemba, ukukhumbula, nezindleko zamaphutha.

Phatha lokhu njengesango lobufakazi: uma imibandela ingafinyelelwa, misa ukukhishwa, vala igebe, bese unweba ukusetshenziswa.

2

Hlola ngedatha efana nezimo zangempela zokukhiqiza.

Phatha lokhu njengesango lobufakazi: uma imibandela ingafinyelelwa, misa ukukhishwa, vala igebe, bese unweba ukusetshenziswa.

3

Engeza isibuyekezo somuntu ukuze uthole ukuzethemba okuphansi noma izibikezelo zomthelela omkhulu.

Phatha lokhu njengesango lobufakazi: uma imibandela ingafinyelelwa, misa ukukhishwa, vala igebe, bese unweba ukusetshenziswa.

4

Landelela ukukhukhuleka kwemodeli bese uqinisekisa kabusha ngemva kwezinguquko zekhamera noma zesethi yedatha.

Phatha lokhu njengesango lobufakazi: uma imibandela ingafinyelelwa, misa ukukhishwa, vala igebe, bese unweba ukusetshenziswa.

Qhubeka Uhlole

I-Computer Vision

Qonda amasistimu ayisisekelo anika amandla i-AI ebonakalayo.

Funda Umhlahlandlela

I-AI Image Generation

Hlola ukugeleza komsebenzi wokudala kanye nokuhwebelana kwamamodeli.

Funda Umhlahlandlela

Check your understanding

Test yourself: take the Score-Based Generative Models quiz

Start quiz →

Related guides