optimization - 随机爬山算法与首选爬山算法

Question

随机爬山算法和首选爬山算法有什么区别？

Answer 1

Hill Climbing Search Algorithm 是本地搜​​索家族中的一种，它根据邻居的更好状态进行移动。Stochastic Hill Climbing 从邻居的所有更好状态中随机选择一个更好的状态，而首选 Hill Climbing 从随机生成的邻居中选择第一个更好的状态。

如果当前状态有很多邻居，首选爬山将成为一个很好的策略。

Answer 2

我引用了 诺维格罗素的人工智能：现代方法（第 3 版）（2010 年）

随机爬山从上坡动作中随机选择；选择的概率会随着上坡移动的陡峭程度而变化。这通常比最陡峭的上升收敛得更慢，但在某些州的情况下，它会找到更好的解决方案。首选爬山通过随机生成后继者，直到生成比当前状态更好的后继者来实现随机爬山。当一个州有许多（例如，数千个）继任者时，这是一个很好的策略。

所以首选爬山是一种特殊的随机爬山。

Answer 3

General hill climbing is a local search algorithm which chooses the best of the neighbor that is it chooses a neighbor with the steepest path and the best objective function value. But due to this it may fail to reach the global maximum and get stuck at the local maximum. Whereas, in the case of stochastic hill climbing it chooses the neighbor with uphill move randomly and the probability of selection might change with the steepness of the uphill moves.In the case of first choice hill climbing it generates the next move randomly and does the search until a state which is better than all the states is found.