InoteveraGaidhi rinotevera
Pseudo-Labeling uye Kuzvidzidzisa
Tekinoroji
Nhungamiro yehunyanzvi
BYOL uye Non-Contrastive Self-Supervision
BYOL (Bootstrap Your Own Latent) inodzidza mifananidzo inobatsira isina mavara uye, zvinoshamisa, pasina mienzaniso yakaipa.
Teerera kugwaro iriAI-yakarondedzerwa · ~2 min
Pa peji ino2 min verenga
Pfupiso
Yakaratidza kuti kudzidza kwekuzvitarisira hakufanire kuvimba nekubvisa mifananidzo yakasiyana, zvichisiya kudiwa kwezvikwata zvakakura zvezvisina kunaka.
Kudzika Kwakadzika
Nzira dzakawanda dzekutanga dzekuzvitarisira dzaive dzakasiyana: vakadhonza maonero maviri emufananidzo mumwechete pamwe chete vachisundidzira mifananidzo yakasiyana-siyana, izvo zvaida masampuli akawanda asina kunaka kudzivirira kuparara (apo network inoburitsa vector imwechete yezvese). BYOL, kubva kuDeepMind muna 2020, yakabvisa zvisina kunaka zvachose. Inoshandisa ma network maviri: online network uye inotarirwa network. Maonero maviri akawedzerwa echifananidzo chimwe chete anoenda nemanetiweki maviri; iyo online network inowedzera musoro wekufanotaura uye inodzidziswa kufanotaura inomiririrwa netiweki yeimwe maonero. Nekutsoropodza, huremu hwetiweki hunonangwa hauna kudzidziswa nekudzika kwe gradient. Panzvimbo iyoyo iwo ari exponential inofamba avhareji (EMA) yehuremu hwepamhepo. Iyi asymmetry pamwe neiyo EMA tarisiro inodzivirira kudonha kushoma nzira dzakasiyana-siyana dzinotyiwa, kuenzanisa kana kurova mutsauko wekutanga paImageNet.
Kuitwa muNyika yeChokwadi
Kufanogadzirira chiono chemusana pamamiriyoni emifananidzo isina kunyorwa, wozonyatsogadzirisa pane diki rakanyorwa remedical-imaging dataset uko kushomeka kwenyanzvi.
Kudzidzira kuona marobhoti zvinhu kubva kuhova dzekamera mbishi pasina kunyora nemaoko, kuderedza mutengo wekudzidzisa mabasa ekugadzirisa.
Kuvaka masisitimu ekudzoreredza uye ekudzoreredza mifananidzo uchishandisa BYOL embeddings inounganidza mifananidzo yakafanana isina mavara ekirasi.
Kutanga setiraiti kana mifananidzo yemuchadenga panzvimbo dzakakura dzisina kunyorwa zvinyorwa tisati tanyatsogadzirisa kushandiswa kwevhu kana kupatsanurwa kwemasango.
Ramangwana reBYOL uye Non-Contrastive Self-Supervision
Pfungwa dzisiri-dzinopikisa zvino dzinotsigira zvakawanda zvekuzvitarisira kuona. SimSiam yakabvisa BYOL pasi zvakare, kuratidza iyo EMA tarisiro hainyanyi kudikanwa kana kumira-gradient ichichengetwa, inodzika kunzwisisa kwechikonzero nei kudonha kuchidziviswa. Tarisira aya mabikirwo ekutanga ekudzidzira asina mavara-emahara kuti arambe akabatana neakafukidzwa-image modelling uye multimodal kudzidziswa, uye kupararira kune vhidhiyo, odhiyo, yekurapa imaging, uye marobhoti uko mavara ari kushomeka kana kudhura, kazhinji senhanho yekudzidzira isati yakareruka inotariswa kugadzirisa zvakanaka.
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 BYOL and Non-Contrastive Self-Supervision 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
Chii chinonzi BYOL uye Non-Contrastive Self-Supervision?
BYOL (Bootstrap Your Own Latent) inodzidza mifananidzo inobatsira isina mavara uye, zvinoshamisa, pasina mienzaniso yakaipa. Yakaratidza kuti kudzidza kwekuzvitarisisa hakufanire kuvimba nekubvisa mifananidzo yakasiyana, zvichisiya kudiwa kwemabheji makuru ezvakaipa.
What makes BYOL non-contrastive?
The BYOL paper highlights learning representations without negative pairs.
What does the online network learn to predict?
BYOL trains the online branch to predict the target representation across two views.
How are target-network weights updated?
The paper describes updating the target with a slow-moving average.
Why does BYOL use an online predictor and stop-gradient on the target side?
The BYOL method uses asymmetry and a stop-gradient as parts of its design.
What happens to two views of the same image in BYOL?
The paper learns by predicting one view’s target representation from another.
Ramba uchidzidza
Mazano anoenderana
Mamwe madhairekitori akasarudzirwa nyaya iyi