[2P5-GS-3] Data mining: Fundamental theory

[2P5-GS-3-04] Ideal State Estimation using Conditional Variational Autoencoder

〇Yuichi Miyamura¹, Tomotake Kozu¹ (1. Deloitte Touche Tohmatsu LLC)

Keywords:deep learning, CVAE, interpretability

In this paper, we report the results of bank marketing campaign analysis using Conditional Variational Autoencoder(CVAE). CVAE is a deep generative model and is often used for image generation. By using this model, we generate the feature vectors of bank customers who subscribe a term deposit in the campaign from the feature vectors of bank customers who do not subscribe it. The difference between the generated feature vectors and the input feature vectors is used for how to improve the campaign.