Xiga xigaHagaha xiga
Calaamadaynta beenta ah iyo is-tababarka
Farsamo
HAGAHA Farsamada
BYOL iyo Kormeerka aan is-hortaagi karin
BYOL (Bootstrap Your Own Latent) waxay bartaa muuqaalada waxtarka leh iyada oo aan wax calaamado ah lahayn iyo, la yaab leh, iyada oo aan lahayn tusaalooyin taban.
Dhageyso hagahanAI-waa laga sheekeeyay · ~2 min
Boggaan2 daqiiqo akhri
Dulmar
Waxay muujisay in waxbarashada is-maamulka ahi aanay u baahnayn in ay kala fogaato muuqaallada aan la midka ahayn, iyada oo dhinac maraysa baahida loo qabo qaybo badan oo taban.
quusid qoto dheer
Inta badan hababka hore ee is-ilaalinta waxay ahaayeen kuwo isbarbardhigaya: waxay soo jiidayeen laba aragtiyood oo la kordhiyey oo isku mid ah iyagoo riixaya sawirro kala duwan, taas oo u baahan muunado badan oo taban si looga fogaado burbur (halkaas oo shabakadu ay soo saarto wax kasta oo isku mid ah). BYOL, oo ka yimid DeepMind sanadka 2020, ayaa gabi ahaanba meesha ka saartay waxyaabaha taban. Waxay isticmaashaa laba shabakadood: shabakad online ah iyo shabakad bartilmaameed ah. Laba aragti oo la kordhiyay oo hal sawir ah ayaa dhex mara labada shabakadood; Shabakadda internetka waxay ku darsataa madaxa saadaasha waxaana loo tababaray inay saadaaliso shabakada bartilmaameedka ah ee matalaadda aragtida kale. Muhiimad ahaan, miisaanka shabakadda bartilmaameedka laguma tababarin faracii hore. Taa beddelkeeda waxay yihiin celceliska dhaqdhaqaaqa jibbaaran (EMA) ee miisaanka onlaynka. Asymmetry-gan oo lagu daray bartilmaameedka EMA waxay ka hortagtaa burburka yar ee hababka isbarbardhigga ah ee laga cabsado, isbarbardhigga ama garaaca saldhigyada isbarbardhigga ee ImageNet.
Dhaqangelinta Adduunka-dhabta ah
Kahortababaridda laf dhabarta aragga malaayiin sawiro ah oo aan calaamadeysan, ka dibna si wanagsan u habeynta xog-sawir caafimaad oo yar oo calaamadeysan halkaas oo faafaahinta khabiiradu ay ku yar yihiin.
Barashada aragtida robot-ka muuqaalada durdurrada kamarada ceeriin iyada oo aan gacanta lagu calaamadin, hoos u dhigida qiimaha baridda hawlaha wax-is-dabamarinta.
Dhisidda sawir-qaadista iyo nidaamyada kala-saarista iyadoo la adeegsanayo isku-xidhka BYOL ee kooxeeya sawiro la mid ah oo aan lahayn calaamado fasal.
Bilaabida moodooyinka sawirka hawada sare ee dayax gacmeedka ama kaydka aan la calaamadin ka hor inta aan si fiican loo hagaajin isticmaalka dhulka ama kala saarista dhirta.
Mustaqbalka BYOL iyo Kormeerka aan is-hortaagi karin
Fikradaha aan is-khilaafsaneyn ayaa hadda ku xidhan aragti is-maamulid badan. SimSiam ayaa hoos u sii dhigay BYOL, tusinaya bartilmaameedka EMA aan si adag loogu baahnayn haddii joogsiga joogsiga la hayo, taasoo sii kordhinaysa fahamka sababta burburka looga fogaado. Ka filo in cuntooyinkan hore ee aan sumadda lahayn ay ku sii biiraan qaabaynta sawirka waji-xidhan iyo tababbarka qaab-dhismeedka kala duwan, iyo in lagu faafiyo muuqaal, maqal, sawir-qaadis caafimaad, iyo qalab-robotik halkaas oo calaamaduhu ku yar yihiin ama qaali yihiin, inta badan sida marxaladda tababbarka ka hor miisaanka khafiifka ah ee kormeerka hagaajinta.
Ku dhaqangelinta
Maxay muhiim u tahay
- Qiimaha iyo miisaaniyada. Go'aamada qaab-dhismeedku waxay horseedaan waxqabadka iyo kharashka hawlgalka sannadaha.
- Go'aamo cad. Waxbarashada farsamada waxay ka caawisaa kooxaha inay doortaan xidhmo sax ah, ma aha oo kaliya kan ugu cusub.
- Xakamaynta tayada. Doorashooyinka injineernimada ee wanaagsan waxay yareeyaan shilalka la isku halleyn karo ee wax soo saarka.
Iska ilaali
- Hagaajinta hal bartilmaameed waxay qarin kartaa daciifnimada nidaamka ballaaran.
- Kaabayaasha dhaqaalaha iyo dayactirka inta badan waa la dhayalsadaa.
- Nabadgelyada iyo daldaloolada u fiirsashada ayaa kori kara marka nidaamyadu noqdaan kuwo aad u adag.
Tallaabooyinka xiga
- Qeex daahida, tayada, iyo bartilmaameedyada qiimaha ka hor inta aan la hirgelin.
- Benchmark marka la eego culeyska dhabta ah iyo xaaladaha xogta.
- La socodka qalabka khaladaadka, leexashada, iyo saamaynta isticmaalaha.
- U diyaari dib-u-noqoshada iyo dariiqyada jawaab-celinta dhacdada ka hor inta aanad miisaan.
Sii wad Sahaminta
Warqad bilaash ah
Hel warbixinta maalinlaha ah ee AI
Saddex sheeko oo la xaqiijiyay oo AI ah maalin kasta oo shaqo ah, oo lagu qoray Ingiriis fudud. Bilaash had iyo jeer, aan xayaysiis lahayn.
Hal iimayl maalin kasta oo shaqo ah. Ka bax hal gujin. Marna ma iibinno ama ma wadaagno ciwaankaaga.
Is tijaabi
Take the BYOL and Non-Contrastive Self-Supervision quiz
Fikrad degdeg ah oo ku saabsan jawaab kasta, iyo shahaado la wadaajin karo oo leh aqoonsi la xaqiijin karo markaad ka gudubto koorso.
Taageer waxbarashada AI bilaashka ah. AI Understanding waa 501(c)(3) aan faa'iido doon ahayn — xayaysiis la'aan, albaab lacag ah ma jiro, had iyo jeer. Samee deeq
Su'aalaha soo noqnoqda
Waa maxay BYOL iyo Kormeerka aan is-hortaagi karin?
BYOL (Bootstrap Your Own Latent) waxay bartaa muuqaalada waxtarka leh iyada oo aan wax calaamado ah lahayn iyo, la yaab leh, iyada oo aan lahayn tusaalooyin taban. Waxay muujisay in waxbarashada is-maamulka ahi aanay u baahnayn in ay kala fogaato muuqaallada aan la midka ahayn, iyada oo dhinac maraysa baahida loo qabo qaybo badan oo taban.
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.
Sii wad waxbarashada
Tilmaamaha la xidhiidha
Tilmaamayaal badan ayaa loo doortay mawduucan