Kyungwoo Song, Deep Ideal Point Estimation with Network, Master's Thesis, Department of Industrial and Systems Engineering, KAIST, 2017

Abstract

For the past few years, probabilistic graphical model (PGM) and deep learning have got the limelight for individual merits. PGM is easy to incorporate prior information or modeler’s assumption and deep learning shows a remarkable performances. Because of this, there are many papers and research about advanced PGM and deep learning lately. Some of the researches are combination of PGM and deep learning such as collaborative deep learning model. In this paper, we proposed a new bayesian deep learning model DIPEN which focuses on legislative system analysis. DIPEN shows the best performance on prediction of legislator’s voting on bill. Further, DIPEN can be used to analyze the legislator’s ideal point, the network between legislators and the ratio of influence between bill and network unlike the previous model which only can be used as an interpretation of individual’s ideal points. 

@masterthesis{Song:2017,

author = {Kyungwoo Song},

advisor ={Il-Chul Moon},

title = {Deep Ideal Point Estimation with Network},

school = {KAIST},

year = {2017}

}