Arabic Sentiment Classification Using Convolutional Neural Network and Differential Evolution Algorithm

الملخص EN

In recent years, convolutional neural network (CNN) has attracted considerable attention since its impressive performance in various applications, such as Arabic sentence classification.

However, building a powerful CNN for Arabic sentiment classification can be highly complicated and time consuming.

In this paper, we address this problem by combining differential evolution (DE) algorithm and CNN, where DE algorithm is used to automatically search the optimal configuration including CNN architecture and network parameters.

In order to achieve the goal, five CNN parameters are searched by the DE algorithm which include convolution filter sizes that control the CNN architecture, number of filters per convolution filter size (NFCS), number of neurons in fully connected (FC) layer, initialization mode, and dropout rate.

In addition, the effect of the mutation and crossover operators in DE algorithm were investigated.

The performance of the proposed framework DE-CNN is evaluated on five Arabic sentiment datasets.

Experiments’ results show that DE-CNN has higher accuracy and is less time consuming than the state-of-the-art algorithms.