We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Approximate Bayesian Computation of Bézier Simplices

Abstract: B\'ezier simplex fitting algorithms have been recently proposed to approximate the Pareto set/front of multi-objective continuous optimization problems. These new methods have shown to be successful at approximating various shapes of Pareto sets/fronts when sample points exactly lie on the Pareto set/front. However, if the sample points scatter away from the Pareto set/front, those methods often likely suffer from over-fitting. To overcome this issue, in this paper, we extend the B\'ezier simplex model to a probabilistic one and propose a new learning algorithm of it, which falls into the framework of approximate Bayesian computation (ABC) based on the Wasserstein distance. We also study the convergence property of the Wasserstein ABC algorithm. An extensive experimental evaluation on publicly available problem instances shows that the new algorithm converges on a finite sample. Moreover, it outperforms the deterministic fitting methods on noisy instances.
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Report number: RIKEN-iTHEMS-Report-21
Cite as: arXiv:2104.04679 [cs.LG]
  (or arXiv:2104.04679v2 [cs.LG] for this version)

Submission history

From: Ken Kobayashi [view email]
[v1] Sat, 10 Apr 2021 04:20:19 GMT (659kb,D)
[v2] Tue, 13 Apr 2021 01:44:44 GMT (660kb,D)

Link back to: arXiv, form interface, contact.