quiz bu lëkkale ak tegtal · Diggante Niveau

Reseau autoroute yi ak laaj lëkkaloo yi

Saytu gaawaayu gated ak residual, yamaleg debit gradient, inspirasioŋ LSTM, jëfandikoo transformatër ak U-Net, ak doxalinu buntu nul-kontra-benn.

Yooni njiit yu jëm ciReseau autoroute yi ak sànni lëkkaloo yi
Laaj 1 de 8

A very deep network struggles with optimization. What can a skip path provide?

Xoolaatal laaj bu nekk8 laaj

Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.

  1. A very deep network struggles with optimization. What can a skip path provide?
    • A guarantee of perfect gradients at every depth
    • A more direct signal and gradient route that can ease training
    • A proof that more layers always improve accuracy
    • A replacement for the loss function
  2. Luy xayma ab bloku residual bu yomb?
    • génne = F(x) yokk ci x
    • génne = x xaaj ak F(x)
    • sorti = li gëna mag ci F(x) ak x
    • génn = F(x) yokk x, fu x nekk gaawaayu dugg
  3. Lan moo inspiré mecanisme gating bi ci reso otoroute yi?
    • Buntu LSTM yu bawoo ci reso yuy baaxoo
    • pooling konvolusioŋ
    • Normalisasioŋ ci lots
    • Dàddu
  4. Ci biir reso otorout, lan la buntu transformaasioŋ T(x) di doxal?
    • Tolluwaayu jàng bu optimisatër bi
    • Dayo nataal biy génn
    • Ñaata ci dugal yi lañuy soppi ak ñaata lañuy yóbbu ci biir
    • Limu couche yi ci reso bi
  5. Ban architecture bu bees moo sukkandiko ci lëkkaloo yi des ci biir ay sous-couche?
    • K-dafay tekki boole
    • Garab giy jël dogal
    • Regresioŋ ligneer
    • Transformatër bi
  6. naka la U-Net mënee jàll lëkkaloo yi gëna jàppale segmentation bi?
    • Dañuy efaase encoder bi yépp
    • Dañuy joxe ay detay yu ndaw ci encodeur bi dem ci decodeur bi
    • Dañuy soppi nataal bi def ko mbind
    • Dañu wàññi limu klaas yiy génne ci benn
  7. In y = F(x)T(x) + x(1-T(x)), what remains when T(x)=0?
    • Only F(x)
    • The sum 2x
    • The carried input x
    • A mandatory zero output
  8. For the same coupled-gate highway form, what remains when T(x)=1?
    • The transformed value F(x)
    • Only the original x
    • F(x) plus x
    • An automatically normalized x