Current browse context:
stat.ML
Change to browse by:
References & Citations
Statistics > Machine Learning
Title: $γ$-ABC: Outlier-Robust Approximate Bayesian Computation Based on a Robust Divergence Estimator
(Submitted on 13 Jun 2020 (v1), last revised 5 Mar 2021 (this version, v3))
Abstract: Approximate Bayesian computation (ABC) is a likelihood-free inference method that has been employed in various applications. However, ABC can be sensitive to outliers if a data discrepancy measure is chosen inappropriately. In this paper, we propose to use a nearest-neighbor-based $\gamma$-divergence estimator as a data discrepancy measure. We show that our estimator possesses a suitable theoretical robustness property called the redescending property. In addition, our estimator enjoys various desirable properties such as high flexibility, asymptotic unbiasedness, almost sure convergence, and linear-time computational complexity. Through experiments, we demonstrate that our method achieves significantly higher robustness than existing discrepancy measures.
Submission history
From: Masahiro Fujisawa [view email][v1] Sat, 13 Jun 2020 06:09:27 GMT (3556kb,D)
[v2] Thu, 22 Oct 2020 08:06:41 GMT (2066kb,D)
[v3] Fri, 5 Mar 2021 05:16:43 GMT (3536kb,D)
Link back to: arXiv, form interface, contact.