Paper Title
CROWDFUNDING PREDICTION WITH ASPECT-BASED CNN AND GENERATIVE ADVERSARIAL NETWORK

Abstract
Abstract - Crowdfunding allows entrepreneurs to raise funds through the Internet. The platforms for crowdfunding give entrepreneurs the chance to raise funds to start entrepreneurial projects. The prosperous development of the Internet has made crowdfunding more and more important. Using machine learning methods to construct a predictive model for forecasting the success of crowdfunding is an important topic in the practical application of financial technology. This research will construct a crowdfunding prediction model based on machine learning and deep learning methods. The proposed model uses the hierarchical attention network to extract the important words and sentence features of campaign descriptions. This research proposes a topic analysis of the latent aspects of campaign descriptions, and constructs a crowdfunding forecast model based on aspect weight analysis. Furthermore, a novel crowdfunding prediction model based on a generative adversarial networks is proposed. The accuracy of the model is increased through adversarial learning of generators and discriminators. The experiments show that our proposed model has better performance. The newly developed model can be applied to crowdfunding platforms in the future. Keywords - Fintech, Crowdfunding, Machine learning, Deep Learning, Latent Topic, Convolutional Neural Network, Generative Adversarial Network