深度学习是机器学习研究的一个新领域,其动机在于建立、模拟人脑进行分析学习的神经网络,模仿人脑的机制来解释数据,是机器学习中一种基于对数据进行表征学习的方法。通过组合低层特征形成更加抽象的高层表示属性类别或特征,以发现数据的分布式特征表示。自2006年Hinton等在深度置信网络方面的重大研究工作发表以来,深度学习作为机器学习的新方向,在人工智能领域的许多重要问题上大显身手。至今已有数种深度学习网络架构,如深度神经网络、卷积神经网络、深度置信网络和递归神经网络,已被成 Deep learning is a new field of machine learning research. Its motivation lies in the establishment of a neural network that simulates the human brain for analytical learning, and a mechanism that mimics the human brain to interpret the data. This is a machine learning method based on characterization of data. A more abstract high-level representation attribute category or feature is formed by combining lower-level features to discover a distributed feature representation of the data. Since Hinton et al. (2006) published a great research work on deep belief networks, deep learning, as the new direction of machine learning, has played an important role in many important issues in the field of artificial intelligence. To date, there have been several deep learning network architectures such as Deep Neural Networks, Convolutional Neural Networks, Deep Confidence Networks and Recurrent Neural Networks that have been