能否介绍一下强化学习（Reinforcement Learning），以及与监督学习的不同？ 第1页

我在外国博客上写过一些基於人工智能的文章，有些搬到了博客园，上面转录的是其中一篇：

我正在研究的 AI architecture 是用强化学习控制 recurrent 神经网络，我相信这个设置可以做到逻辑推理和答问题的功能，基本上就是 strong AI。 但还有一些未解决的细节。 论文的标题是《游荡在思考的迷宫中》，即将发表。

補充： 還有一點，就是監督學習的問題可以很容易化為強化學習的問題（雖然這樣增加了複雜性而沒有益處），但反之則沒有一般的辦法。 見：Reinforcement Learning and its Relationship to Supervised Learning，Barto and Dietterich, 2004.

"But is it possible to do this the other way around: to convert a reinforcement learning task into a supervised learning task?

"In general, there is no way to do this. The key difficulty is that whereas in supervised learning, the goal is to reconstruct the unknown function f that assigns output values y to data points x, in reinforcement learning, the goal is to find the input x* that gives the maximum reward R(x*).

"Nonetheless, is there a way that we could apply ideas from supervised learning to perform reinforcement learning? Suppose, for example, that we are given a set of training examples of the form (xi, R(xi)), where the xi are points and the R(xi) are the corresponding observed rewards. In supervised learning, we would attempt to find a function h that approximates R well. If h were a perfect approximation of R, then we could find x* by applying standard optimization algorithms to h."