Ikibazo gihuza ibibazo · Hagati Urwego

Ikibazo gisanzwe

Reba intego zisanzwe, ibihano L1 na L2, guta ishuri, guhagarara hakiri kare, imbaraga zibihano, nurugero rwumubare wintego isanzwe.

Inzira zijyanye no kuyoboraGusanzwe
Ikibazo 1 Bya 8

Why would a practitioner add regularization when a model fits training-specific noise?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Why would a practitioner add regularization when a model fits training-specific noise?
    • To guarantee zero training error
    • To remove the need for validation data
    • To guarantee a faster GPU kernel
    • To discourage overfitting and evaluate whether unseen-data performance improves
  2. Nubuhe buryo busanzwe bushobora gutwara uburemere kuri zeru, guhitamo neza ibiranga?
    • L2 isanzwe (kubora ibiro)
    • L1
    • Ibisanzwe
    • Kongera igipimo cyo kwiga
  3. Nigute abata ishuri basanzwe bahuza imiyoboro yimitsi mugihe cyamahugurwa?
    • Isiba ibice burundu kurusobe
    • Yongeraho uburemere bwa kwaduka kumikorere yo gutakaza
    • Irazimya kuzimya neuron buri ntambwe kugirango idashobora guhuza
    • Irahagarika imyitozo mugihe igihombo cyo kwemeza kizamutse
  4. Mu gihombo gisanzwe 'ikosa + lambda × ibihano', kwiyongera kwa labda gukora iki?
    • Shimangira ibihano, usunika icyitegererezo kugisubizo cyoroshye
    • Gabanya igihano kandi yemerera uburemere bunini
    • Kuraho ijambo ryamakosa rwose
    • Nta ngaruka igira ku bipimo
  5. Kuki guhagarara hakiri kare bishobora gufatwa nkuburyo busanzwe?
    • Kuberako byongera igihano cyibiro kubihombo
    • Kuberako ihagarika imyitozo mbere yuko moderi itangira guhuza urusaku mumibare
    • Kuberako byongera ubushobozi bwicyitegererezo mugihe
    • Kuberako ikuraho neuron
  6. What can happen when regularization is strengthened beyond what a task needs?
    • Both training and test accuracy must improve
    • The penalty becomes unrelated to the objective
    • Validation becomes unnecessary
    • The model can underfit useful structure even if variance decreases
  7. For a regularization example with weights minus two and one, what are the unscaled L1 and sum-of-squares L2 penalties?
    • L1 is three and L2 is five
    • L1 is minus one and L2 is three
    • L1 is five and L2 is three
    • Both penalties are minus two
  8. A regularized objective has prediction error two, lambda one-half, and penalty six. What is its total value?
    • Umunani
    • Bitatu
    • Bitanu
    • Imwe