中文摘要：

深度学习是机器学习领域的研究热点,它使机器学习更加接近人工智能。作为深度学习的一类经典模型,卷积神经网络已被广泛应用于语音识别、图像识别和自然语言处理等领域中。本文探讨了卷积神经网络的基本原理、实现及应用。首先回顾了卷积神经网络的发展历史,阐述了它的基本原理,研究了卷积层和下采样层;其次总结了卷积神经网络的三大重要特性：稀疏连接、权值共享和池采样,并将卷积神经网络应用在MNIST手写体数字识别任务中;最后指出了卷积神经网络未来的重点研究方向。

英文摘要：

Deep learning is a new research focus in the field of machine learning and its emergence makes machine learning closer to the goal of artificial intelligence. A s a classical model in deep learn-ing, convolution neural network has been widely applied in the fields of speech recognition, image rec-ognition, natural language processing, etc. This paper discussed the basic principle, realization and applications of convolution neural network. Firstly, it reviewed the history of convolution neural net-work and elaborated its basic principle of convolution neural network and investigated the convolution layer and the sub - sampling layer. Secondly, it summarized the three important characteristics of con-volution neural network, i.e., sparse connection, weight sharing and sub - sampling. The handwrit-ten digits recognition task of convolution neural network was also realized in the MNIST database. Finally, it gave future key research directions for convolution neural network.