A GAN-based Transfer Learning Approach for Sentiment Analysis

Abstract

Transfer learning is an important artificial intelligence approach which extracts knowledge from source domain to solve tasks in the target domain. As a research hot topic, Generative Adversarial Networks (GAN) provides a powerful framework in constructing unsupervised models. A GAN consists of two neural networks: a discriminator to distinguish natural and generated samples, and a generator to deceive the discriminator. Generally, sentiment analysis of text is a big challenge in the Natural Language Processing (NLP). In this paper, A GAN-based transfer learning approach for sentiment analysis of cross-domain texts is presented. Experiments are performed in cross-domain e-commerce reviews. The results compiled demonstrate the effectiveness of the proposed approach.