摘要
自2006年Hinton提出新的神经网络学习算法以来,卷积神经网络(CNN)在图像识别、语音识别等领域取得巨大成功,引发一股深度学习热潮。伴随着基于海量文本的词向量表示技术的进步,卷积神经网络逐渐在自然语言处理领域取得超越传统机器学习算法的识别效果。综述CNN在关系抽取领域的最新研究成果,介绍其发展历程、基本框架,并对CNN在该领域的发展方向进行总结和展望。
Since 2006, Hinton proposed a new neural network learning algorithm, CNN (Convolutional Neural Network) in image recognition, speech recognition and other areas of great success, triggering a deep learning boom. Convolutional neural networks have gradually gained recognition beyond the traditional machine learning algorithms in the field of natural language processing, along with the advances in word vector representation technology based on massive text. Reviews CNN's latest research achievements in the field of relational extraction, introduces its development process and basic framework, and summarizes the development direction of CNN in this field.
出处
《现代计算机》
2017年第6期58-60,共3页
Modern Computer
关键词
卷积神经网络
词向量
关系抽取
Convolutional
Word Vector Representation
Relation Extraction