Gaussian Neuron in Deep Belief Network for Sentiment Prediction

Deep learning has been widely applied in natural language processing. The neuron model in a deep belief network is important for its performance, and so more attention should be paid to investigate how much influence the neuron will play on its results. In this paper we investigate the neuron's effect for sentiment prediction, and then apply both total accuracy and F-measure to evaluate the performance. Finally, our experimental results show the idea of Gaussian neuron performs relatively better on the Stanford Twitter Sentiment corpus, which further proves the neuron model should be considered for a specific problem.