Kedis ku xidhan hagaha · Dhexdhexaad Heerka

Kediska Saadaasha waafaqsan

Wuxuu hubiyaa saadaasha waafaqsan saadaasha, beddelka, buundooyinka cabbirka, xadka tirada iyo fikradaha ka dambeeya caynsanaanta-muunad-koobeedka ah.

Wadooyinka haga ee la xidhiidhaSaadaasha waafaqsan
Su'aal 1 ee 8

Maxaa soo saaraya habka saadaasha waafaqsan halkii hal summad?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Maxaa soo saaraya habka saadaasha waafaqsan halkii hal summad?
    • Saadaasha go'an ama dhexda oo leh dammaanad caymis
    • Qaab dib loo tababaray oo saxnaan sare leh
    • Hal dhibic oo qiyaas ah oo aan la hubin
    • Qiimaha luminta tababarka
  2. Male-awaalkee ayaa xoojinaya dammaanadda caymiska caadiga ah?
    • Qaabku waa shabakad neural ah
    • Dhibcaha xogta waa la isweydaarsan karaa
    • Tilmaamuhu waa ka madaxbannaan yihiin midba midka kale
    • Calaamaduhu si fiican ayay isu dheeli tiran yihiin
  3. Waa maxay doorka jaangooynta?
    • Si loo tababaro qaabka hoose
    • Si loo doorto heerka waxbarashada
    • Si loo xisaabiyo buundooyinka aan waafaqsanayn ee dejinaya heerka
    • Si meesha looga saaro kuwa ka baxsan xogta tababarka
  4. Haddii aad dejiso alfa = 0.1, caymiskee ayaad beegsanaysaa?
    • 10%
    • 50%
    • 99%
    • 90%
  5. Maxaa caadi ahaan ku dhacaya saadaasha la dejiyay marka moodelku aanu aad u hubin wax gelinta?
    • Qalabku wuu weynaadaa, oo ka kooban calaamado murashaxiin badan
    • Qalabka wuxuu noqonayaa mid madhan
    • Meesha waa la iska indhatiray
    • Qaabka si toos ah ayaa dib loogu tababaray
  6. Waa maxay sababta saadaasha waafaqsan loogu yeedhay 'qaybin-free'?
    • Waxay kaliya ku shaqeysaa xogta Gaussian
    • Uma baahnid mala-awaal gooni ah oo ku saabsan qaybinta xogta dammaanad qaadkeeda
    • Waxay meesha ka saaraysaa qaybinta xogta gebi ahaanba
    • Waxay u malaynaysaa qaybin isku mid ah
  7. A split-conformal classifier has 99 calibration examples and targets 90% marginal coverage. Which rank is used for the calibration quantile?
    • The 10th ordered score
    • The 99th score only
    • The 90th ordered calibration score (ceil((99+1) × 0.9))
    • The average of all labels
  8. In classification, what can a nonconformity score such as 1 minus the candidate label probability represent?
    • Tirada waayihii tababarka
    • The probability that the calibration set is exchangeable
    • The class frequency in the test set
    • How poorly the model supports that candidate label for the example