Ikibazo gihuza ibibazo · Hagati Urwego

Imiyoboro Yumuhanda no Gusiba Ikibazo

Gerageza amarembo asigaranye, ibisigisigi-bigenda, imipaka ya LSTM, guhinduranya na U-Net ikoresha, hamwe na zeru-na-imyitwarire imwe.

Inzira zijyanye no kuyoboraImiyoboro Yumuhanda Kandi Simbuka
Ikibazo 1 Bya 8

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

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  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. Ni ubuhe buryo bwibanze busigaye bubara?
    • ibisohoka = F (x) byikubye x
    • ibisohoka = x bigabanijwe na F (x)
    • ibisohoka = ntarengwa ya F (x) na x
    • ibisohoka = F (x) wongeyeho x, aho x niyinjiza ryihuta
  3. Niki cyahumekeye uburyo bwo kwinjirira mumihanda minini?
    • Amarembo ya LSTM kuva kumurongo usubiramo
    • Guhuriza hamwe
    • Ibisanzwe
    • Kureka
  4. Mumuyoboro munini, irembo ryo guhindura T (x) rigenzura iki?
    • Igipimo cyo kwiga cya optimizer
    • Ingano y'ibisohoka
    • Ni bangahe byinjira byahinduwe bitandukanye bitwarwa neza
    • Umubare wibice murusobe
  5. Ni ubuhe bwubatsi bugezweho bushingiye kumasigarira asigaye hafi ya sublayers?
    • K-bisobanura guhuriza hamwe
    • Igiti cyemezo
    • Gusubira inyuma
    • Impinduka
  6. Nigute U-Net gusimbuka guhuza bifasha cyane cyane igice?
    • Basiba kodegisi yose
    • Batambutsa neza ahantu hatandukanye kuva kodegisi kugeza kuri decoder
    • Bahindura ishusho kumyandiko
    • Bagabanya umubare wibisohoka ibyiciro kuri imwe
  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