Deep Forest as a framework for a new class of machine-learning models

• 来源期刊: 国家科学评论（英文版）
• 年，卷（期）: 2019,（2）
• 页码范围: 186-187
• 页数: 2页
• 语种: 英文
• DOI: 10.1093/nsr/nwy151

摘要

A new deep learning framework — the socalled Deep Forest (DF), proposed by Zhi-Hua Zhou andJi Feng [1,2]-can be regarded as one of the important events of 2017 in machine learning, although it was unjustly unnoticed by a large number of researchers.The DF combines several ensemble-based methods, including Random Forests (RFs) and Stacking, into a structure that is similar to a multi-layer neural network, but each layer in the DF contains RFs instead of neurons.All advantages of DF are dearly discussed in [1-3].In particular, DF is simple for training due to a very small number of hyper-parameters, it does not use backpropagation training and it outperforms many well-known methods, including deep neural networks, when there are only small-scale training data.