Repository logo

Infoscience

  • English
  • French
Log In
Logo EPFL, École polytechnique fédérale de Lausanne

Infoscience

  • English
  • French
Log In
  1. Home
  2. Academic and Research Output
  3. Conferences, Workshops, Symposiums, and Seminars
  4. Safe non-smooth black-box optimization with application to policy search
 
conference paper

Safe non-smooth black-box optimization with application to policy search

Usmanova, Ilnura
•
Krause, Andreas
•
Kamgarpour, Maryam  
July 31, 2020
Proceedings of the 2nd Conference on Learning for Dynamics and Control
Learning for Dynamics and Control

For safety-critical black-box optimization tasks, observations of the constraints and the objective are often noisy and available only for the feasible points. We propose an approach based on log barriers to find a local solution of a non-convex non-smooth black-box optimization problem minf0(x)minf0(x)\min f^0(x) subject to fi(x)≤0,i=1,…,mfi(x)≤0,i=1,…,mf^i(x)\leq 0, i = 1,\ldots, m, at the same time, guaranteeing constraint satisfaction while learning with high probability. Our proposed algorithm exploits noisy observations to iteratively improve on an initial safe point until convergence. We derive the convergence rate and prove safety of our algorithm. We demonstrate its performance in an application to an iterative control design problem.

  • Details
  • Metrics
Type
conference paper
Author(s)
Usmanova, Ilnura
Krause, Andreas
Kamgarpour, Maryam  
Date Issued

2020-07-31

Publisher

PMLR

Published in
Proceedings of the 2nd Conference on Learning for Dynamics and Control
Start page

980

End page

989

URL
https://proceedings.mlr.press/v120/usmanova20a.html
Editorial or Peer reviewed

REVIEWED

Written at

OTHER

EPFL units
SYCAMORE  
Event nameEvent date
Learning for Dynamics and Control

2020-07-31

Available on Infoscience
December 1, 2021
Use this identifier to reference this record
https://infoscience.epfl.ch/handle/20.500.14299/183437
Logo EPFL, École polytechnique fédérale de Lausanne
  • Contact
  • infoscience@epfl.ch

  • Follow us on Facebook
  • Follow us on Instagram
  • Follow us on LinkedIn
  • Follow us on X
  • Follow us on Youtube
AccessibilityLegal noticePrivacy policyCookie settingsEnd User AgreementGet helpFeedback

Infoscience is a service managed and provided by the Library and IT Services of EPFL. © EPFL, tous droits réservés