A0835
Title: Piecewise monotone estimation in one-parameter exponential families
Authors: Takeru Matsuda - RIKEN Center for Brain Science (Japan)
Yuto Miyatake - Osaka University (Japan) [presenting]
Abstract: The problem of estimating a piecewise monotone sequence of normal means is called the nearly isotonic regression. An efficient algorithm has been devised for this problem by modifying the pool adjacent violators algorithm (PAVA). We are concerned with estimating a piecewise monotone sequence for general one-parameter exponential families such as binomial, Poisson, and chi-square. We propose an efficient algorithm based on the modified PAVA, which utilizes the duality between the natural and expectation parameters. We also provide a method for selecting the regularization parameter using an information criterion. Simulation results demonstrate that the proposed method detects change-points in piecewise monotone parameter sequences in a data-driven manner. We present several applications such as spectrum estimation, causal inference and discretization error quantification of ODE solvers.
