“∇-Prox: Differentiable Proximal Algorithm Modeling for Large-scale Optimization” by Lai, Wei, Fu, Härtel and Heide
Conference:
Type(s):
Title:
- ∇-Prox: Differentiable Proximal Algorithm Modeling for Large-scale Optimization
Session/Category Title: Surfaces, Strips, Lights
Presenter(s)/Author(s):
Moderator(s):
Abstract:
∇-Prox is a domain-specific language and compiler for large-scale optimization that transforms optimization problems into differentiable proximal solvers. With a few lines of code, ∇-Prox makes it easier to prototype different learning-based, bi-level optimization algorithms based on user requirements with regard to memory constraints or training budget for a diverse range of applications.
Additional Images:
![©Zeqiang Lai, Kaixuan Wei, Ying Fu, Philipp Härtel, and Felix Heide](https://history.siggraph.org/wp-content/uploads/2024/02/2023-Tech-Papers-Lai_∇-Prox-Differentiable-Proximal-Algorithm-Modeling-for-Large-Scale-Optimization.jpg)