Presentation information

International Session

International Session » ES-2 Machine learning

[2S5-IS-2c] Machine learning

Wed. Jun 15, 2022 3:20 PM - 5:00 PM Room S (Online S)

Chair: Jun Sakuma (University of Tsukuba)

4:40 PM - 5:00 PM

[2S5-IS-2c-05] Overfitting Problem in the Approximate Bayesian Computation Method Based on Maxima Weighted Isolation Kernel

Overfitting Problem in the ABC Method Based on MaxW iKernel

〇Iurii S. Nagornov1 (1. The National Institute of Advanced Industrial Science and Technology)


Keywords:Overfitting problem, Approximate Bayesian computation, Isolation Kernel, Maxima weighted isolation kernel mapping

Recently we designed the heuristic approximate Bayesian computation method based on maxima weighted isolation kernel. The method showed good results on parameter estimation for the branching processes model that has unevenly stochastically distributed data. This work is devoted to the problem of the fitting and overfitting of the internal parameters for the method such as a number of Voronoi sites and a number of trees of isolation forest. Here, we discuss the reasons for overfitting and how to fit parameters properly with an example of a two-dimension task.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.