2021-05-05 10:13:49 -07:00
|
|
|
---
|
|
|
|
|
id: 5e8f2f13c4cdbe86b5c72da4
|
2021-07-16 11:03:16 +05:30
|
|
|
title: '使用 Q-Learning 進行強化學習：第 2 部分'
|
2021-05-05 10:13:49 -07:00
|
|
|
challengeType: 11
|
|
|
|
|
videoId: DX7hJuaUZ7o
|
2021-10-03 12:24:27 -07:00
|
|
|
bilibiliIds:
|
|
|
|
|
aid: 420570359
|
|
|
|
|
bvid: BV1G341127zr
|
|
|
|
|
cid: 409139190
|
2021-05-05 10:13:49 -07:00
|
|
|
dashedName: reinforcement-learning-with-q-learning-part-2
|
|
|
|
|
---
|
|
|
|
|
|
|
|
|
|
# --question--
|
|
|
|
|
|
|
|
|
|
## --text--
|
|
|
|
|
|
2021-07-16 11:03:16 +05:30
|
|
|
如果智能體在採取隨機動作和使用學習動作之間沒有很好的平衡，會發生什麼？
|
2021-05-05 10:13:49 -07:00
|
|
|
|
|
|
|
|
## --answers--
|
|
|
|
|
|
2021-07-16 11:03:16 +05:30
|
|
|
智能體將始終嘗試將其對當前狀態/動作的獎勵最小化，從而導致局部最小值。
|
2021-05-05 10:13:49 -07:00
|
|
|
|
|
|
|
|
---
|
|
|
|
|
|
2021-07-16 11:03:16 +05:30
|
|
|
智能體將始終嘗試將其對當前狀態/動作的獎勵最大化，從而導致局部最大值。
|
2021-05-05 10:13:49 -07:00
|
|
|
|
|
|
|
|
## --video-solution--
|
|
|
|
|
|
|
|
|
|
2
|
|
|
|
|
|
