InoteveraGaidhi rinotevera
Federated Learning
Tekinoroji
Nhungamiro yehunyanzvi
Contrastive Learning
Kudzidza kwakasiyana kunodzidzisa modhi kudhonza zvinhu zvakafanana pamwe chete uye kusundira zvinhu zvakasiyana munzvimbo yekumisikidza.
Teerera kugwaro iriAI-yakarondedzerwa · ~2 min
Pa peji ino2 min verenga
Pfupiso
It matters because it lets AI learn powerful representations from mostly unlabeled data, powering image search, recommendations, and multimodal models.
Kudzika Kwakadzika
Panzvimbo pekufanotaura zita, kudzidza kwakasiyana kunodzidza nekuenzanisa: kupihwa chinhu cheancho, modhi inodzidziswa kuitira kuti inofananidzwa 'yakanaka' imhare pedyo nayo munzvimbo yevekita nepo isingaenderane 'negative' ivhu kure. Recipe yakajairwa yekuzvitarisira (seSimCLR) inogadzira zvakanaka nekutora maviri akasarudzika ekuwedzera emufananidzo mumwechete (chirimwa, ruvara jitter, blur); zvimwe zvese zviri mubatch hazvina kunaka. Iyo modhi inoisa mamepu kune mavekita uye kurasikirwa kunopa mubairo wepamusoro kufanana kwevaviri uye kuderera kufanana kwevamwe. Izvi zvinogadzira zvinomisikidza uko chinhambwe chinoratidza zvinoreva, saka basa rezasi rinoda mavara mashoma. CLIP inoshandisa iyo pfungwa imwe chete mumodalities, ichifananidza mifananidzo kune zvinyorwa zvavo.
Kuitwa muNyika yeChokwadi
CLIP kudzidza nzvimbo yakagovaniswa yemifananidzo-yemavara kuitira kuti utsvage muraibhurari yemifananidzo ine mutsara wakanyorwa senge 'imbwa iri pabhodhi rekutamba'.
Kufanogadzirira chiono chemusana neSimCLR pamifananidzo isina kunyorwa, wozoigadzirisa kuti ionekwe nechirwere iine kadiki kakanyorwa chete.
Kuvaka chigadzirwa kana nziyo kurudziro uko zvakamisikidzwa kwezvinhu zvakafarirwa nemushandisi zvinogara padhuze kuti zvitore muvakidzani ari pedyo.
Masisitimu ekusimbisa kumeso anodzidzisa kuisirwa saka mafoto maviri emunhu mumwe ari padyo uye vanhu vakasiyana vari kure.
Ramangwana reKudzidza Kunopesana
Kudzidza kwakasiyana kuri kusangana neakafukidzwa uye anogadzira kuzvitarisira kuita zvinangwa zvakasanganiswa zvinotora kufanana kwepasirese uye ruzivo rwakanaka. Mhedzisiro yaro yakakura ndeye multimodal: inomisikidzwa yakamisikidzwa mufananidzo-chinyorwa (uye ikozvino odhiyo nevhidhiyo) embeddings inotsigira kutsvaga, kudzoreredza-yakawedzera chizvarwa, uye zero-pfuti kupatsanura, uye iwo tsoka inokura. Tarisira basa rakawanda mukuderedza havi yezvimedu zvakakura, pakuwedzera kuri nani uye nzira dzisina kunaka dzekuchera migodhi, uye pakuwedzera nzira yekuenda kumatunhu akaita sekufungidzira kwekurapa uye nguva-yakatevedzana uko mavara ari kushomeka uye anodhura.
Kuzvishandisa
Nei zvichikosha
- Mutengo uye bhajeti. Zvisarudzo zvezvivakwa zvinotyaira kuita uye mutengo wekushandisa kwemakore.
- Sarudzo dzakajeka. Dzidzo yehunyanzvi inobatsira zvikwata kusarudza murwi wakakodzera, kwete iwo mutsva chete.
- Kudzora kwemhando yepamusoro. Sarudzo dzeinjiniya dziri nani dzinoderedza zviitiko zvekuvimbika mukugadzira.
Chenjera
- Kugadzirisa imwe bhenji kunogona kuvanza yakafara system kushaya simba.
- Infrastructure uye mari yekugadzirisa inowanzotarisirwa pasi.
- Chengetedzo uye kucherechedzwa mapundu anogona kukura sezvo masisitimu anowedzera kuoma.
Matanho anotevera
- Tsanangura latency, mhando, uye mutengo zvinangwa usati waitwa.
- Benchmark pasi pechokwadi mutoro uye data mamiriro.
- Chishandiso chekutarisa zvikanganiso, kudonha, uye mushandisi maitiro.
- Gadzirira nzira dzekudzosera kumashure uye dzezviitiko usati wawedzera.
Ramba Uchiongorora
Tsamba yemahara
Tora iyo yemazuva ese AI pfupiso
Nhatu dzakasimbiswa dzeAI nyaya mangwanani ega ega pakati pevhiki, dzakanyorwa neChirungu chakajeka. Yemahara zvachose, hapana ads.
Imwe email zuva rega rega revhiki. Bvisa kunyoresa kamwe chete. Hatimbotengese kana kugovera kero yako.
Zviedze iwe pachako
Take the Contrastive Learning quiz
Mhinduro yepakarepo pamhinduro yega yega, uye chitupa chinogovaniswa chine ID inobatika kana wapedza kosi.
Tsigira dzidzo yemahara yeAI. AI Understanding ndeye 501(c)(3) isingabatsiri - hapana ads, hapana paywall, zvachose. Ita chipo
Mibvunzo inowanzo bvunzwa
What is Contrastive Learning?
Kudzidza kwakasiyana kunodzidzisa modhi kudhonza zvinhu zvakafanana pamwe chete uye kusundira zvinhu zvakasiyana munzvimbo yekumisikidza. Izvo zvine basa nekuti inoita kuti AI idzidze inomiririra ine simba kubva kuzhinji isina kunyorwa dhata, simba rekutsvaga mifananidzo, kurudziro, uye multimodal modhi.
Ndechipi chinangwa chikuru chekudzidza kwakasiyana?
Kudzidza kwakasiyana kunoumba nzvimbo yekumisikidza kuitira kuti mapeya anoenderana ari padyo uye asingaenderane mapeya ari kure kure.
Munzira yekuzvitarisira yemufananidzo seSimCLR, mbiri yakanaka inogadzirwa sei?
SimCLR inoumba zvakanaka kubva kune maviri akawedzera maonero echifananidzo chimwe, nemimwe mifananidzo mubatch inoshanda seyakaipa.
Nderipi basa rekurasikirwa rinonyanya kuenderana nekudzidza kwakasiyana?
InfoNCE, softmax pamusoro pekufanana zvibodzwa zvine tembiricha, ndiyo yakajairika inosiyanisa chinangwa.
Sei nzira dzakaita seMoCo dzichishandisa memory bank kana mutsara?
Kudzidza kwakasiyana kunobatsira kubva kune zvakawanda zvakashata; mutsara unovapa ivo vachichengeta batch size inogoneka.
Idambudziko ripi iro BYOL neSimSiam vanonyatso dzivirira pasina kushandisa zvakajeka zvisiri izvo?
Pasina zvakaipa, modhi inogona kuita mepu zvese kune imwechete vector; kumira-gradient uye nhanho dzekusimudzira zvinodzivirira kudonha uku.
Ramba uchidzidza
Mazano anoenderana
Mamwe madhairekitori akasarudzirwa nyaya iyi