InoteveraGaidhi rinotevera
Kuyera Mitemo yeNeural Networks
Basics
Basics GUIDE
Neural Tangent Kernel Theory
Iyo Neural Tangent Kernel (NTK) chishandiso chemasvomhu chinoratidza kuti neural network yakafara kwazvo inoita seyakatarwa, yakagadziriswa kernel nzira panguva yekudzidziswa.
Teerera kugwaro iriAI-yakarondedzerwa · ~2 min
Pa peji ino2 min verenga
Pfupiso
It matters because it turns mysterious deep learning into something with closed-form, analyzable equations.
Kudzika Kwakadzika
Yakaunzwa naJacot, Gabriel, naHongler muna 2018, NTK dzidziso inodzidza zvinoitika sematanho etiweki anova akafara zvisingaite. Mumuganho iwoyo, kudzidziswa ne gradient descent kunomira kuve rwendo rwemusango rusina mutsara: maparamendi etiweki haafambi kubva payakagadzika kutanga (iyo 'usimbe kudzidzira' hutongi), uye basa rainokokorodza rinoshanduka rakaita mutsara, richitongwa nekernel inoramba iripo panguva yese yekudzidziswa. Iyo kernel - yemukati chigadzirwa chegradient nekuremekedza ma paramita - ndiyo NTK. Nekuti kernel regression ine mhinduro chaidzo, unogona kufanotaura kuburitswa kwenetiweki yakadzidziswa pasina kuidzidzisa. NTK yakatsanangura kuti nei yakakura overparametered network ichigona kukwana data asi ichiri kuita zvakajairika, uye inobatanidza kudzidza kwakadzama kumakumi emakumi emakore ezvinonzwisisika kernel nzira uye Gaussian maitiro.
Kuitwa muNyika yeChokwadi
Kufanotaura netiweki yakakura yekudzidzira dynamics nekuongorora kusarudza mareti ekudzidza pasina anodhura bvunzo inomhanya
Kushandisa NTK-based metrics kukwirisa mumiriri zvivakwa zvakachipa panguva yekutsvaga neural architecture
Kutsanangura ne theoretically kuti nei overparametered network ichichinja kusvika zero kudzidziswa kurasikirwa uye ichiri kuita generalize
Kugadzira kernel approximations (NTK-yakafuridzirwa Gaussian maitiro) emabasa ane diki dhata uko chaiko kusavimbika fungidziro ine basa.
Ramangwana reNeural Tangent Kernel Theory
NTK ndiyo musana wedzidziso yemazuva ano yakadzama-yekudzidza, asi chaiyo inogumira network inodzidzira maficha - chimwe chinhu chinorasikirwa nemufananidzo wekernel. Tsvagiridzo ikozvino inotarisa pamukaha uripo pakati pe'usimbe' NTK maitiro uye 'kupfuma' maitiro-yekudzidza hurumende, uye pakushandisa NTK kufanotaura mashandiro ezvivakwa, kutungamira neural architecture yekutsvaga, uye yakasungwa generalization. Tarisira dzidziso dzakasanganiswa dzinobata kana network ichiita senge kernels maringe kana vachidzidza zvinomiririra.
Kuzvishandisa
Nei zvichikosha
- Sarudzo dzakajeka. Inokubatsira kuparadzanisa zvakajeka zvichemo zvehunyanzvi kubva mumutauro wekushambadzira.
- Mutengo uye bhajeti. Iwe unogona kubvunza zvirinani kuita mibvunzo usati washandisa mari kana nguva.
- Team uye workflow. Zvikwata zvine nzwisiso yakagovaniswa inoita zvirinani chigadzirwa, mutemo, uye sarudzo dzekudzidza.
Chenjera
- Zvikwata zvakasiyana zvinogona kushandisa izwi rimwechete zvakasiyana, saka tsanangura nzvimbo nekukurumidza.
- Benchmarks inogona kutaridzika yakasimba nepo chaiyo-yenyika kuita isina kuenzana.
- Kuregeredza mhando yedata uye zvirongwa zvekuongorora zvinowanzogadzira mhedzisiro isina kusimba.
Matanho anotevera
- Tanga netsanangudzo yemutauro wakajeka yemhedzisiro yaunoda.
- Sarudza metric imwe yekubudirira uye imwe yekutadza mamiriro usati waedzwa.
- Mhanya mutyairi mudiki ane data remumiriri, kwete demo rakakwenenzverwa.
- Gwaro uko Neural Tangent Kernel Theory inobatsira uye uko nzira dzakareruka dziri nani.
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 Neural Tangent Kernel Theory 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 Neural Tangent Kernel Theory?
Iyo Neural Tangent Kernel (NTK) chishandiso chemasvomhu chinoratidza kuti neural network yakafara kwazvo inoita seyakatarwa, yakagadziriswa kernel nzira panguva yekudzidziswa. Izvo zvine basa nekuti inoshandura isinganzwisisike yakadzika kudzidza kuita chimwe chinhu chine yakavharwa-fomu, inoongororwa equations.
Mune isingaperi-hupamhi muganho, iyo Neural Tangent Kernel inoita sei panguva yekudzidziswa?
Mhedzisiro yepakati NTK mhedzisiro ndeyekuti mune isingaperi-hupamhi muganho kernel inochinjika kune yakagadziriswa kukosha uye inoramba yakagadzikana mukati me gradient-kudzika kudzidziswa.
Chii chinonzi NTK nemasvomhu?
Iyo NTK ndeye K(x,x') = ⟨∇θ f(x), ∇θ f(x')⟩, chigadzirwa chemukati chemagetsi ezvinobuda maererano nemaparamita.
Ndiani akaunza Neural Tangent Kernel?
Iyo NTK yakaunzwa mubepa re2018 naArthur Jacot, Franck Gabriel, naClément Hongler.
Chii chinonzi 'usimbe kudzidzisa' hutongi?
Mukudzidzira usimbe, maparamendi e-network anogara padyo nemaitiro avo ekutanga uye zvinobuda zvinoshanduka zvine mutsetse, izvo zvinoita kuti iyo yakagadziriswa-kernel ongororo ishande.
Nekuti iyo NTK inoderedza kudzidziswa kune kernel regression, chii chinogoneka?
Kernel regression ine yakavharwa-fomu mhinduro, saka mumuganho weNTK munhu anogona kufembera nekuongorora zvakabuda netiweki yakadzidziswa.
Ramba uchidzidza
Mazano anoenderana
Mamwe madhairekitori akasarudzirwa nyaya iyi