Ikibazo gihuza ibibazo · Hagati Urwego

Ikibazo-Kwiga Ikibazo

Kugenzura Q-indangagaciro, kwiga hanze ya politiki, kuvugurura Bellman, gushakisha, guhuza ibitekerezo hamwe nimpamvu Q-kwiga ari moderi-yubusa.

Inzira zijyanye no kuyoboraQ Kwiga
Ikibazo 1 Bya 8

Niki Q-agaciro Q (s, a) igereranya?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Niki Q-agaciro Q (s, a) igereranya?
    • Igihembo giteganijwe igihe kirekire cyo gufata ingamba muri leta s, hanyuma ugakora neza
    • Amahirwe yo kuba muri leta s
    • Inshuro inshuro ibikorwa a byageragejwe
    • Igihembo cyihuse kubikorwa a
  2. Kuki Q-Kwiga bisobanurwa ngo 'hanze ya politiki'?
    • Irasaba icyitegererezo cyibidukikije
    • Irashobora kwiga politiki nziza mugihe ukurikiza politiki itandukanye, yubushakashatsi
    • Ikora kumurongo gusa kumibare yabitswe
    • Irengagiza ibihembo rwose
  3. Mu mategeko yo kuvugurura, ni iki kintu cyo kugabanya gamma igenzura?
    • Igipimo cyubushakashatsi
    • Ingano yimbonerahamwe ya leta
    • Ni bangahe ibihembo bizaza bifite uburemere ugereranije nibyihuse
    • Umubare wibice
  4. Ni irihe kosa ry'agateganyo (TD) muri Q-Kwiga?
    • Itandukaniro hagati ya leta ebyiri
    • Igihembo cyose cyigice
    • Igipimo cyo kwiga inshuro gamma
    • r + gamma * max Q (s ', a') ukuyemo ikigereranyo cya Q (s, a) igereranyo
  5. Ni ukubera iki imbonerahamwe isanzwe Q-Kwiga irwana nibibazo binini nkimikino ya videwo kuva kuri pigiseli?
    • Umubare wibihugu bitandukanye ni munini cyane kubika kumeza
    • Ntishobora gukoresha ibihembo
    • Ikora gusa nibikorwa bibiri
    • Gamma igomba kunganya 1
  6. Niyihe ntego yo guhitamo ibikorwa bya epsilon-umururumba?
    • Guhitamo buri gihe Q-agaciro ibikorwa
    • Kuringaniza gushakisha ibikorwa bishya hamwe no gukoresha ibyiza bizwi
    • Gushiraho kugabanuka
    • Kubika Q-Imbonerahamwe
  7. What exploration condition appears in the tabular Q-learning convergence result?
    • Only the greedy action is ever sampled
    • Each state must be visited exactly once
    • Every state-action pair must be visited repeatedly under suitable learning conditions
    • The transition model must be known in advance
  8. Why is Q-learning called model-free?
    • It never observes the environment
    • It does not use rewards
    • It requires a complete transition table before learning
    • It updates values from sampled rewards and next states without storing transition probabilities