Directed Subdifferentiable Functions and the Directed Subdifferential without Delta-Convex Structure
Robert Baier, Elza Farkhi, Vera Roshchina

TL;DR
This paper introduces a new way to construct the directed subdifferential for a broader class of functions using only directional derivatives, removing the need for delta-convex structure, thus expanding its applicability.
Contribution
It generalizes the directed subdifferential to functions beyond delta-convex, including Lipschitz and quasidifferentiable functions, using a novel construction from directional derivatives.
Findings
Directed subdifferential constructed without delta-convex structure
Applicable to Lipschitz functions definable on o-minimal structures
Extends to quasidifferentiable functions
Abstract
We show that the directed subdifferential introduced for differences of convex (delta-convex, DC) functions by Baier and Farkhi can be constructed from the directional derivative without using any information on the DC structure of the function. The new definition extends to a more general class of functions, which includes Lipschitz functions definable on o-minimal structure and quasidifferentiable functions.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
