20 citations · 23 across the 7 of their papers we have counts for
7 papers · 1 filter
A Unified Model for High-Resolution ODEs: New Insights on Accelerated Methods
Hoomaan Maskan, Konstantinos C. Zygalakis, Armin Eftekhari +1
Recent work on high-resolution ordinary differential equations (HR-ODEs) captures fine nuances among different momentum-based optimization methods, leading to accurate theoretical…
Principal Component Hierarchy for Sparse Quadratic Programs
Robbie Vreugdenhil, Viet Anh Nguyen, Armin Eftekhari +1
We propose a novel approximation hierarchy for cardinality-constrained, convex quadratic programs that exploits the rank-dominating eigenvectors of the quadratic matrix. Each level…
Training Linear Neural Networks: Non-Local Convergence and Complexity Results
Armin Eftekhari
Linear networks provide valuable insights into the workings of neural networks in general. This paper identifies conditions under which the gradient flow provably trains a linear n…
PCA by Optimisation of Symmetric Functions has no Spurious Local Optima
Raphael A. Hauser, Armin Eftekhari
Principal Component Analysis (PCA) finds the best linear representation of data, and is an indispensable tool in many learning and inference tasks. Classically, principal component…
Explicit Stabilised Gradient Descent for Faster Strongly Convex Optimisation
Armin Eftekhari, Bart Vandereycken, Gilles Vilmart +1
This paper introduces the Runge-Kutta Chebyshev descent method (RKCD) for strongly convex optimisation problems. This new algorithm is based on explicit stabilised integrators for…
Sparse Inverse Problems Over Measures: Equivalence of the Conditional Gradient and Exchange Methods
Armin Eftekhari, Andrew Thompson
We study an optimization program over nonnegative Borel measures that encourages sparsity in its solution. Efficient solvers for this program are in increasing demand, as it arises…