Find Optimal Model Among Various Neural Networks Models Using Monte-Carlo Tree Search
Jaeyoung Moon, Misun Kim, Dongil Han · 2018
Nowadays, most researchers in machine learning field will agree that the deep neural networks(DNNs) provide the best performance in pattern recognition, computer vision, natural language processing and so on. So many DNNs architectures are built relying on mathematical heuristics and prior knowledge. We can make some new models by ourselves depending on the platforms, the type of data and applications. If we can experiment all possible cases, then we will find the best approximate solution. Unfortunately, it is hard to do like that because experimenting every possible case costs too much resources. In this paper, we suggest the method to find optimal neural networks model among various models by using Monte-Carlo Tree Search [1].