基于BiGRU和贝叶斯分类器的文本分类

针对传统的循环神经网络模型在处理长期依赖问题时面临着梯度爆炸或者梯度消失的问题,且参数多训练模型时间长,提出一种基于双向GRU神经网络和贝叶斯分类器的文本分类方法。利用双向GRU神经网络提取文本特征,通过TF-IDF算法权重赋值,采用贝叶斯分类器判别分类,改进单向GRU对后文依赖性不足的缺点,减少参数,缩短模型的训练时间,提高文本分类效率。在两类文本数据上进行对比仿真实验,实验结果表明,该分类算法与传统的循环神经网络相比能够有效提高文本分类的效率和准确率。

Text classification based on bidirectional GRU and Bayesian classifier

Aiming at the problem that the traditional recurrent neural network model faces the gradient explosion or gradient disappearance when dealing with long-term dependence problems,and that the parameter scale is large and training model takes a long time,a text classification method based on two-way GRU neural network and Bayesian classifier was proposed.The two-dimensional GRU neural network was used to extract the text features,the weights were assigned using the TF-IDF algorithm,and the Bayesian classifier was used to discriminate the classification.The shortcomings of the one-way GRUs are improved,and the parameter scale is reduced.The training time of the model is greatly reduced,and the efficiency of text classification is improved.By comparing experiments on two types of text data,the experimental results show that the classification algorithm can effectively improve the efficiency and accuracy of text classification compared with traditional cyclic neural networks.