Idanwo ti o sopọ mọ itọsọna · Alabọde Ipele

Idanwo Ẹkọ Imudara Aisinipo

Explore fixed-dataset reinforcement learning, distribution shift, and representative conservative and sequence-modeling methods.

Jẹmọ itọsọna awọn ọnaẸkọ Imudara Aisinipo
Ibeere 1 ti 8

Kini o ṣe asọye ikẹkọ aisinipo (ipele) ikẹkọ iranlọwọ?

Awotẹlẹ gbogbo ibeere8 ibeere

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

  1. Kini o ṣe asọye ikẹkọ aisinipo (ipele) ikẹkọ iranlọwọ?
    • Kọ ẹkọ lati inu data ti o wa titi laisi ibaraenisepo agbegbe tuntun
    • Kọ ẹkọ nikan nipasẹ iṣawari ifiwe nigbagbogbo
    • Ikẹkọ laisi eyikeyi ifihan agbara ere
    • Nṣiṣẹ inference lai eyikeyi awoṣe
  2. Kini ipenija imọ-ẹrọ aringbungbun ni RL offline?
    • Aini ti eyikeyi data
    • Iyipada pinpin ati overestimation ti jade-ti-pinpin sise
    • Iwakiri ori ayelujara ti o pọju
    • Awọn paramita diẹ diẹ ninu nẹtiwọọki
  3. What does D4RL provide to researchers evaluating offline reinforcement-learning algorithms?
    • A guarantee that an offline policy is safe to deploy
    • A new optimizer that removes the need for logged data
    • A benchmark suite of datasets and tasks for offline RL evaluation
    • A dataset containing only online interactions collected by each test policy
  4. Bawo ni Konsafetifu Q-Learning (CQL) dojuko overestimation?
    • Nipa gbigbe awọn iye fun awọn iṣe ti a ko rii
    • Nipa yiyọ ere naa kuro patapata
    • Nipa ṣawari diẹ sii ni agbaye gidi
    • Nipa ṣiṣe deede awọn iye-Q si isalẹ fun awọn iṣe pinpin-jade
  5. Bawo ni Amunawa Ipinnu ṣe atunṣe RL aisinipo?
    • Gẹgẹbi awoṣe ti o tẹle ti o sọ asọtẹlẹ awọn iṣe ti o ni ilodi si ipadabọ ti o fẹ
    • Bi iṣupọ ti awọn ipinlẹ
    • Bi funfun ID search
    • Bi abojuto aworan classification
  6. Kilode ti awọn algoridimu RL ori ayelujara ti o ṣe deede ko ṣee lo bi-jẹ offline?
    • Won ni ko si nkankikan nẹtiwọki
    • Wọn ko le gba data atunṣe lati ṣatunṣe iwọn apọju iye
    • Wọn ko lo awọn ere
    • Wọn ko beere eto imulo kankan
  7. How does BCQ reduce reliance on actions absent from an offline dataset?
    • It adds random actions to every training transition
    • It assumes every action has been observed equally often
    • It constrains candidate actions to stay near actions the dataset supports
    • It removes the action-value function and reward
  8. What value-learning method is central to the cited Implicit Q-Learning paper?
    • Maximum-likelihood fitting of a language tokenizer
    • Uniform regression over actions sampled from a live environment
    • A supervised image-classification loss
    • Expectile regression over values associated with dataset actions