An Introduction to Recurrent Neural Networks and their Application in Sentiment Analysis

Abstract

Over the years Artificial Neural Networks have become an incredibly powerful tool, finding many varied applications in the modern world. Artificial Neural Networks, which are modeled on the structure of the human brain, have the ability to generate complex inferences from data that was previously impossible with traditional algorithms. With the exponential increase in the amount and complexity of data, Artificial Neural Networks are becoming integral to data analytics. However, traditional Neural Networks do not have the ability to effectively process sequential data, such as text or video. This is one of the main drawbacks of traditional Feed Forward Neural Networks. In this paper, we will discuss the Recurrent Neural Network architecture and how its design leads to its unique ability to deal with sequential data and address the drawback of traditional Feed-Forward Neural Networks. We will then perform a comparative analysis of many different types of Recurrent Neural Networks and display their effectiveness with a standard problem from Natural Language Processing: Sentiment analysis.