Bayesian generalized fused lasso modeling via NEG distribution

Abstract

The fused lasso penalizes a loss function by the L₁ norm for both the regression coefficients and their successive differences to encourage sparsity of both. In this paper, we propose a Bayesian generalized fused lasso modeling based on a normal-exponential-gamma (NEG) prior distribution. The NEG prior is assumed into the difference of successive regression coefficients. The proposed method enables us to construct a more versatile sparse model than the ordinary fused lasso using a flexible regularization term. Simulation studies and real data analyses show that the proposed method has superior performance to the ordinary fused lasso.

Keywords:

• Bayesian lasso
• Bayesian model
• hierarchical normal-exponential-gamma distribution
• Markov chain Monte Carlo

Acknowledgement

The computational resource was provided by the Super Computer System, Human Genome Center, Institute of Medical Science, The University of Tokyo.

Additional information

Funding

M. Ueki was supported by Grant-in-Aid for Young Scientist (B) (25870074) and Grants-in-Aid for Scientific Research (C) (25330049 and 25460403). S. Kawano was supported by Grant-in-Aid for Young Scientist (B) (15K15947) and Grants-in-Aid for Scientific Research on Innovative Areas (16H06429, 16K21723, and 16H06430).