题目
COMP90054_2025_SM1 Exam: AI Planning for Autonomy (COMP90054_2025_SM1) - Requires Respondus LockDown Browser
单项选择题
In a single iteration of the Monte-Carlo Tree Search (MCTS) algorithm, what is the primary purpose of the "Simulate" step (also known as rollout)?
选项
A.To estimate the long-term value (expected future reward) from a newly expanded state by playing out a random or heuristic-guided sequence of actions until a terminal state is reached.
B.To select the most promising unexpanded node in the current search tree using a multi-armed bandit strategy like UCB.
C.To add new child nodes to a selected leaf node, representing possible outcomes of an unexplored action.
D.To update the visit counts and Q-values of all ancestor nodes in the path from the root to the newly expanded node, based on the outcome of the simulation.
查看解析
标准答案
Please login to view
思路分析
Examining the Monte-Carlo Tree Search (MCTS) pipeline, we focus on what the Simulate (rollout) phase does in a single iteration.
Option 1: 'To estimate the long-term value (expected future reward) from a newly expanded state by playing out a random or heuristic-guided sequence of actions until a terminal state ......Login to view full explanation登录即可查看完整答案
我们收录了全球超50000道考试原题与详细解析,现在登录,立即获得答案。
更多留学生实用工具
希望你的学习变得更简单
加入我们,立即解锁 海量真题 与 独家解析,让复习快人一步!