What are the current models of deep learning and what problems are they applicable to

AlphaGo relies on an accurate expert evaluation system (value network):An expert system is an intelligent computer program system that contains a large amount of knowledge and experience at the level of an expert in a particular field, and is able to utilize the knowledge and problem-solving methods of human experts to deal with problems in that field.

Deep neural network (policy network) based on massive data:The advantage of multilayer is that complex functions can be represented with fewer parameters. In supervised learning, the problem with previous multilayer neural networks was that they tended to fall into local extrema. If the training samples are sufficient to adequately cover future samples, then the learned multilayer weights can be well used to predict new test samples. But it is difficult to get enough labeled samples for many tasks, in which case simple models such as linear regression or decision trees often give better results than multilayer neural networks. In unsupervised learning, there has been no effective way to construct multilayer networks in the past. The top layer of a multi-layer neural network is a high-level representation of the underlying features, e.g., the bottom layer is pixel dots, and the nodes in the upper layer may represent horizontal lines, triangles; while the top layer may have a node representing a face.

Traditional Artificial Intelligence Methods Monte Carlo Tree Search for Combinatorial:is a method for making optimal decisions in artificial intelligence problems, generally in the form of action (move) planning in combinatorial games. It combines the generality of stochastic simulation with the accuracy of tree search.