Ikibazo gihuza ibibazo · Hagati Urwego

Ikibazo cyo Kurwanya Imiterere

Inkomoko yikizamini hamwe nintego ihindagurika, ihinduranya indangarubuga ya gradients, intego-label ubuke, uburyo butagira isoko, hamwe nintego zitandukanye.

Inzira zijyanye no kuyoboraKurwanya Imiterere
Ikibazo 1 Bya 8

Ni ikihe kibazo kijyanye no guhuza imiterere yagenewe gukemura?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Ni ikihe kibazo kijyanye no guhuza imiterere yagenewe gukemura?
    • Kugabanya umubare wibipimo byicyitegererezo
    • Kwihutisha kugwiza matrix kuri GPUs
    • Kugabanuka kwimikorere mugihe amakuru yikizamini atandukanye namakuru yamahugurwa
    • Kubyara imyitozo ngororamubiri ibirango byikora
  2. Muri domaine-adversarial net, niki icyiciro cya reversion layer ikora?
    • Irasunika ibiranga gukuramo kugirango ikore isoko nintego itamenyekana
    • Isiba indangarubuga mugihe cyo gufata umwanzuro
    • Bikubye kabiri igipimo cyo kwiga kumahugurwa yihuse
    • Ihindura ibisubizo byo gusubira inyuma mubirango byashyizwe mubikorwa
  3. Niki gituma 'kutagenzurwa' guhuza n'imihindagurikire y'ikirere bigira akamaro cyane?
    • Ntabwo bisaba imyitozo yicyitegererezo na gato
    • Ikora gusa kumashusho yamakuru
    • Ikuraho ibikenewe kumurongo winkomoko
    • Ntabwo ikeneye amakuru yanditseho indangarubuga
  4. Ni ibihe bintu ni urugero rwiza rwo guhinduranya indangarubuga?
    • Icyitegererezo cyatojwe kandi cyageragejwe kuri dataset imwe
    • Ikibyimba cyamahugurwa yatojwe kuri scaneri imwe yananiwe kuri scaneri itandukanye
    • Icyitegererezo gikora vuba nyuma yo kubara
    • Umuyoboro ufite ibice byinshi kurenza urundi rusobe
  5. Ni ubuhe buryo 'butagira isoko' uburyo bwo kurwanya imihindagurikire y'ikirere bugenewe kwirinda?
    • Kugera kumasoko yumwimerere yamahugurwa mugihe cyo kurwanya
    • Gukoresha imiyoboro iyo ari yo yose yubatswe
    • Gutanga ibisohoka byose
    • Gukorera kuri GPU zigezweho
  6. A team starts adaptation from a broadly pretrained model. What should it still measure?
    • Only the parameter count
    • Only source training accuracy
    • Performance on the actual target data and task
    • Whether target evaluation can be skipped
  7. Which DANN head supplies the objective to distinguish source examples from target examples?
    • The task label vocabulary
    • The domain classifier
    • The optimizer checkpoint file
    • A fixed output embedding dimension
  8. Does gradient reversal mean changing a positive class label into a negative class label?
    • Yes; every task label is negated
    • Yes; all target samples are relabeled as source
    • No; it freezes every feature weight
    • No; it reverses the domain gradient into the feature extractor