Wasserstein GAN
Wasserstein GAN (WGAN) waa dib-u-qaabaynta ujeedada tababarka GAN ee adeegsata masaafada Wasserstein halkii luminta min-max asalka ah.
Dulmar
It makes notoriously unstable GAN training far more reliable and gives a loss value that actually correlates with image quality.
quusid qoto dheer
GAN-yada asalka ah waxay tababaraan laba shabakadood oo isku-jiidjiid ah: koronto-dhaliye wuxuu sameeyaa sawirro been abuur ah iyo takoorid ayaa isku dayaya inuu arko. Tani inta badan way burburtaa ama istaagtaa sababtoo ah khasaaraha takoorku ma sheego wax faa'iido leh oo ku saabsan horumarka. WGAN, oo ay soo bandhigeen Arjovsky, Chintala, iyo Bottou sanadka 2017, waxay ku bedeshay takoorida 'dhaleec' kaas oo keenaya sida dhabta ah ee sawirku ugu ekaado miisaan joogto ah halkii la kala saari lahaa dhabta vs-fake. Bartilmaameedka tababarku wuxuu noqdaa masaafada u dhaxaysa Wasserstein (dhaqdhaqaaqa dhulka) ee u dhexeeya qaybinta xogta dhabta ah iyo tan la sameeyay. Fogaantani waxay siinaysaa jaanisyo fudud oo macno badan xitaa marka labada qaybintu ay si dhib yar isu dulsaaran yihiin, taasoo hoos u dhigaysa burburka qaabka iyo samaynta luminta qalooca calaamad tayo leh oo dhab ah.
Aragtida Farsamada
Fogaanta Wasserstein waxay si dareen leh u cabbirtaa 'shaqada' ugu yar si ay hal tuulo oo wasakh ah (qaybinta beenta ah) u rogaan mid kale (ta dhabta ah). Xisaabintu waxa ay ku xidhan tahay labada qaybood ee Kantorovich-Rubinstein, taas oo u baahan in naqdiyehu noqdo 1-Lipschitz (xajmiga xadaysan). WGAN-kii asalka ahaa ayaa tan si xun u dhaqangeliyay isagoo gooyay miisaan yar oo kala duwan; WGAN-GP ka dib wuxuu ku beddelay googooyn ciqaab yar oo si tartiib ah u riixaysa naqdiga dhaleeceynta heerka 1, tababarka si adag.
Saamaynta Istiraatijiyadeed
Xawaaraha iyo miisaanka
Visual AI wuxuu si otomaatig ah u samayn karaa baadhista, ogaanshaha, iyo sumadaynta hawlaha miisaanka.
Xulashada dhismayaasha
Kooxaha hal-abuurka leh waxay hindise karaan fikradaha si dhakhso leh iyagoo leh dib-u-eegis buugeed yar.
Kooxda iyo socodka shaqada
Hawlgalladu waxay isticmaali karaan calaamadaha muuqaalka iyo muuqaalka kuwaas oo markii hore adkeyd in la farsameeyo.
Mustaqbalka Wasserstein GAN
Aragtida asaasiga ah ee WGAN, in doorashada masaafada qaybinta ay qaabayso tayada isjiidka, wali waxay ku soo noqnoqonaysaa qaabaynta wax-soo-saarka. In kasta oo moodooyinka faafintu ay hadda xukumaan isku-darka sawirka, fikradaha gaadiidka ugu habboon ee WGAN ayaa dib ugu soo baxaya is-waafajinta qulqulka, hababka buundada ee Schrodinger, iyo kala-soocidda moodooyinka faafinta oo loo beddelo koronto-dhaliyaha tillaabo yar yar. Filo ujeedooyinka qaabka Wasserstein si ay u sii wargeliyaan hababka isku-dhafka ah ee tababarka xasiloon iyo mitirka khasaaraha macnaha leh, gaar ahaan qaybaha sayniska iyo xogta hoose.
Dhaqangelinta Adduunka-dhabta ah
Soo saarista wajiyo sawir leh iyo muuqaalo muuqaal ah oo vanilj GANs ay ku burbureen wax soo saaryo dhawr ah oo soo noqnoqday
Soo saarida sawiro caafimaad oo la isku daray, sida MRI ama balastar histology, si loo kordhiyo xog-ururinta calaamadeysan.
Ku qaabaynta dhacdooyinka isku dhaca qayb ka mid ah jilitaannada fiisigiska tamarta sare leh halkaasoo tababarka xasillooni uu muhiim u yahay
U adeegidda sidii halbeegga aasaasiga ah ee cilmi-baarista ML sababtoo ah khasaaraheedu waxay raad raacaan muunadda muunada ee tababarka
Khatarta & Dariiqyada Ilaalada
Xuquuqda sawirka iyo ogolaanshaha waxay noqon kartaa khataro sharci ah haddii caddayntu aanay caddayn.
Waxqabadka moodeelku wuu ku kala duwanaan karaa iftiinka, tirakoobka, iyo deegaanka.
Wanaagga beenta ah waxa laga yaabaa inaan la dareemin ilaa xadka kalsoonida aan la kormeerin.
Qorshe Hawleedka Dhaqangelinta
Qeex shuruudaha aqbalida ee saxnaanta, dib u celinta, iyo kharashyada khaladka.
Ku tijaabi xogta ku habboon xaaladaha wax soo saarka dhabta ah.
Ku dar dib u eegis bini'aadamka si aad u hesho kalsoonida hoose ama saameeynta sare.
Lasoco moodeel dhaqaaqa oo dib u cusboonaysii kamarada ama xogta kaydinta ka dib.
Sii wad Sahaminta
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Wasserstein GAN quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Hagaha xiga
ESRGAN iyo GAN Super-Resolution
Su'aalaha soo noqnoqda
What is Wasserstein GAN?
Wasserstein GAN (WGAN) waa dib-u-qaabaynta ujeedada tababarka GAN ee adeegsata masaafada Wasserstein halkii luminta min-max asalka ah. Waxay ka dhigtaa tababbarka GAN aan xasilloonayn caan ka fog oo la isku halleyn karo waxayna siinaysaa qiimo lumis dhab ahaantii xiriir la leh tayada sawirka.
Waa maxay mitir masaafada ay WGAN isticmaasho si ay u barbar dhigto qaybinta dhabta ah iyo tan la soo saaray?
WGAN waxay ku beddeshaa ujeedadii asalka ahayd ee Jensen-Shannon ku salaysan masaafada Wasserstein, taas oo bixisa jaangooyooyin fudud xitaa marka qaybintu ay si dhib yar isku dhaafto.
WGAN, shabakadii ahayd haybsooca ayaa loo beddelay maxaa, maxaa yeelay?
Naqdintu waxay u dhalisaa sawirada miisaan joogto ah halkii ay kala saari lahaayeen dhabta iyo been-abuurka, taas oo ah waxa ka dhigaya ujeedada Wasserstein.
Maxay tahay sababta dhaleeceynta WGAN loogu xaddiday inuu noqdo 1-Lipschitz?
Labadan qaybood ee u oggolaanaya WGAN qiyaasida masaafada Wasserstein waxay haysaa oo keliya shaqooyinka 1-Lipschitz, sidaa darteed jaangooyooyinka dhaleeceynta waa in la xaddidaa.
Sidee buu WGAN-kii asalka ahaa u dhaqan geliyay xannibaadda Lipschitz?
Waraaqdii ugu horreysay ee WGAN waxay isticmaashay jarid miisaan cayriin ah; WGAN-GP ka dib waxay ku bedeshay rigoore jilicsan oo jilicsan.
Dhibaato noocee ah ayay WGAN si gaar ah u yaraynaysaa marka loo eego vaniljka GANs?
Jilayaasha jilicsan ee WGAN waxay xakameeyaan qaabka burburay waxayna soo saaraan khasaare dhab ahaantii la xidhiidha tayada muunada.