Ikibazo gihuza ibibazo · Hagati Urwego

Kwigana Kwiga Ikibazo

Kugenzura cloni yimyitwarire, guhinduranya, kwerekana impuguke, gukusanya amakuru ya DAgger hamwe nimbogamizi hanze yigihugu cyagaragaye.

Inzira zijyanye no kuyoboraKwiga Kwigana
Ikibazo 1 Bya 8

Mu kwigana kwigana, ni gute umukozi abona imyitwarire?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Mu kwigana kwigana, ni gute umukozi abona imyitwarire?
    • Kwiga politiki ukoporora imyigaragambyo
    • Kugwiza igihembo cyateguwe n'intoki binyuze mubushakashatsi butunguranye
    • Gucomeka urusobe rw'imitsi kugirango rukore kubikoresho bito
    • Gukora amashusho avuye mwandiko
  2. Imyitwarire ya cloni yimyitwarire yigana kwigana nkikibazo ki?
    • Ihuriro ridakurikiranwa
    • Kugenzura imyigire ya leta-ku-bikorwa
    • Igisekuru
    • Gushiraho ibihembo
  3. Ni ikihe kibazo gitera cloni yimyitwarire kunanirwa mugihe kirekire?
    • Birenze kubikorwa byo guhemba
    • Kubura gradients muri kodegisi
    • Gukomatanya ikosa kuva kugabura kwimuka
    • Kwibagirwa kwibagirwa ibyiciro bishaje
  4. Nigute algorithm ya DAgger ikemura iyo ntege nke?
    • Mugukuraho umuhanga muri loop rwose
    • Mugukomeza kugabanuka
    • Mugushyiramo urusaku rudasanzwe mubikorwa byose
    • Kubaza impuguke kuri leta uwiga asura kandi akusanya ayo makuru
  5. Ni ukubera iki kwigana kwigana bikundwa kuruta kwiga gushimangira imirimo nko gutwara?
    • Imyiyerekano iroroshye gutanga kuruta ibikorwa byiza byo guhemba
    • Iremeza imyitwarire myiza
    • Ntabwo ikenera amakuru ayo ari yo yose
    • Irakora byihuse kubitekerezo gusa
  6. What is a realistic limitation of behavioral cloning when the agent reaches states outside the demonstrations?
    • It is guaranteed to improve beyond the expert
    • Its behavior is not guaranteed there because those states may be out of distribution
    • It can only produce random actions
    • It automatically queries an expert at every step
  7. In one DAgger iteration, what does the learner do before requesting expert labels?
    • Deletes all earlier demonstrations
    • Trains only on randomly generated labels
    • Rolls out its current policy and records the states it actually visits
    • Replaces the expert policy permanently
  8. After the expert labels the states visited by the current policy, what happens to that data?
    • It is used only as an evaluation metric
    • It replaces the original observation space
    • It is discarded after one gradient step
    • It is aggregated with previous labeled data for retraining