Weak contraction mapping and optimizationDownload PDF

27 Sep 2018 (modified: 21 Dec 2018)ICLR 2019 Conference Blind SubmissionReaders: Everyone
  • Abstract: The weak contraction mapping is a self mapping that the range is always a subset of the domain, which admits a unique fixed-point. The iteration of weak contraction mapping is a Cauchy sequence that yields the unique fixed-point. A gradient-free optimization method as an application of weak contraction mapping is proposed to achieve global minimum convergence. The optimization method is robust to local minima and initial point position.
  • Keywords: Weak contraction mapping, fixed-point theorem, non-convex optimization
  • TL;DR: A gradient-free method is proposed for non-convex optimization problem
8 Replies