Active Deep Networks for Semi-Supervised Sentiment Classification

Experiments show that ADN outperforms the semi-supervised learning algorithm and deep learning techniques applied for sentiment classification.

Abstract

This paper presents a novel semi-supervised learning algorithm called Ac-tive Deep Networks (ADN), to address the semi-supervised sentiment classifica-tion problem with active learning. First, we propose the semi-supervised learning method of ADN. ADN is constructed by Restricted Boltzmann Machines (RBM) with unsupervised learning using labeled data and abundant of unlabeled data. Then the constructed structure is fine-tuned by gradient-descent based super-vised learning with an exponential loss function. Second, we apply active learn-ing in the semi-supervised learning framework to identify reviews that should be labeled as training data. Then ADN architecture is trained by the se-lected labeled data and all unlabeled data. Experiments on five sentiment classifica-tion datasets show that ADN outper-forms the semi-supervised learning algo-rithm and deep learning techniques ap-plied for sentiment classification. 1