Ikibazo gihuza ibibazo · Hagati Urwego

Ikibazo cyo Kureka no Kuringaniza Ikibazo

Gerageza masike idahwitse, imyitozo nuburyo bwerekana, gupima guhindagurika, imipaka yo guhuza n'imihindagurikire y'ikirere, guta MC, no guhanura guhinduka.

Ikibazo 1 Bya 8

Abata ishuri bakora iki mugihe cy'amahugurwa?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Abata ishuri bakora iki mugihe cy'amahugurwa?
    • Gusiba ibiro burundu
    • Mubisanzwe ushyiraho agace ka neuron gasohoka kuri zeru kuri buri cyerekezo cyimbere
    • Kabiri igipimo cyo kwiga
    • Ongeraho izindi nzego kumurongo
  2. Kuki guta ishuri bitezimbere rusange?
    • Yongera umubare wibipimo
    • Bikuraho ibikenewe byamahugurwa
    • Iremeza ikosa rya zeru
    • Irinda neuron gufatanya no guhuza imbaraga, guhagararirwa gukomeye
  3. Bigenda bite guta ishuri mugihe cyibizamini (umwanzuro)?
    • Irazimye kandi umuyoboro wuzuye urakoreshwa
    • Igabanuka ndetse na neuron nyinshi
    • Isiba kimwe cya kabiri cyibiro
    • Itondekanya ibirango
  4. Igipimo cyo guta p = 0.5 murwego bisobanura hafi iki mugihe cyamahugurwa?
    • Neuron zose zavanyweho
    • Nta neuron igira ingaruka
    • Hafi ya kimwe cya kabiri cya neuron zeru kuri buri pass
    • Igice cyasibwe
  5. Ni iki guta ishuri bikunze gusobanurwa nko kugereranya?
    • Gusubira inyuma kumurongo umwe
    • Igiti
    • Impinduka
    • Ihuriro ryimiyoboro myinshi yoroheje igabana uburemere
  6. Niki Monte Carlo yataye ishuri ikoreshwa?
    • Kwihutisha amakuru yipakurura
    • Kugereranya icyitegererezo kidashidikanywaho mugukomeza guta ishuri
    • Gucomeka kumurongo kuburemere buke
    • Kuranga amakuru yamahugurwa mu buryo bwikora
  7. With inverted dropout, why is no extra activation scaling normally needed at ordinary inference?
    • The keep probability becomes exactly zero
    • Inference always averages new random masks
    • Surviving activations were already scaled during training
    • Every weight is divided by the number of layers
  8. The same input yields varying predictions because dropout remains active during evaluation. What explains the variation?
    • New random masks sample different subnetworks on each pass
    • The input changes its label by itself
    • The network has no trained weights
    • Batch size must be one