AI期末复习(PolyU)

文档详见评论链接

Tutorial 2-Search Algorithm
Breadth-first search宽度优先搜索
[图片]
[图片]
open表由queue队列实现，因此加在尾部。
[图片]
Depth-first search深度优先搜索
[图片]
[图片]
open表由stack栈实现，因此加在头部。
[图片]
Hill climbing (a heuristic search algorithm)
[图片]
Hill climbing algorithm goes uphill along the steepest possible path until can go no further up, which may return a state that is a local maximum.
Advantages

Avoid traversal避免遍历所有解。
Achieve the purpose of improving efficiency提高搜索效率。
Disadvantages
Not necesssarily find the global maximum but converge on a local maximum不一定找到最优解但收敛于局部最优解。
In plateau cases, the hill climber may not be able to determine in which direction it should step, and may wander in a direction that never leads to improvement.在高原(比较平坦)的情况下，登山者可能无法确定应该朝哪个方向前进，并且可能在一个永远无法改善的方向上徘徊。
Ridges problem: If the target function creates a narrow ridge that ascends in a non-axis-aligned direction, then the hill climber can only ascend the ridge by zig-zagging.山脊问题:如果目标函数创建了一个狭窄的山脊，它以非轴线对齐的方向上升，那么爬山者只能通过曲折攀登山脊。

Best-first Search (Greedy Search贪心算法)
The node with the lowest evaluation is expanded first, i.e., 𝑎𝑟𝑔𝑚𝑖𝑛(𝑓(𝑛))
f(n) = ℎ(𝑛) = 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝑐𝑜𝑠𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑐ℎ𝑒𝑎𝑝𝑒𝑠𝑡 𝑝𝑎𝑡ℎ 𝑓𝑟𝑜𝑚 𝑠𝑡𝑎𝑡𝑒 𝑎𝑡 𝑛𝑜𝑑𝑒 𝑛 𝑡𝑜 𝑎 𝑔𝑜𝑎𝑙 𝑠𝑡𝑎𝑡𝑒
If 𝑛 is a goal node, then ℎ(𝑛) = 0.
[图片]
Limitations of Greedy Search

not optimal
[图片]
A* Search
[图片]
[图片]
[图片]
[图片]
Exercise
[图片]
Breadth-first search：A B C D E F
[图片]
Depth-first search：A D F
[图片]
[图片]
[图片]
[图片]
[图片]
openList, closeList = [start], []
while True:
currentNode = lowest f cost in openList
if currentNode == end: return
for neighbour in currentNode.Neighbours:
if closeList.contains(neighbour) continue
if new_neighbour_f <= old_neighbour_f or not openList.contains(neighbour):
neighbour.f = new_neighbour_f
if not openList.contains(neighbour):
openList.add(neighbour)
[图片]
[图片]
[图片]
[图片]
[图片]
[图片]
[图片]
[图片]
[图片]
[图片]