9 citations · 15 across the 4 of their papers we have counts for
4 papers · 1 filter
EnKSGD: A Class Of Preconditioned Black Box Optimization And Inversion Algorithms
Brian Irwin, Sebastian Reich
In this paper, we introduce the Ensemble Kalman-Stein Gradient Descent (EnKSGD) class of algorithms. The EnKSGD class of algorithms builds on the ensemble Kalman filter (EnKF) line…
Deep Neural Network Accelerated Implicit Filtering
Brian Irwin, Eldad Haber, Raviv Gal +1
In this paper, we illustrate a novel method for solving optimization problems when derivatives are not explicitly available. We show that combining implicit filtering (IF), an exis…
Secant Penalized BFGS: A Noise Robust Quasi-Newton Method Via Penalizing The Secant Condition
Brian Irwin, Eldad Haber
In this paper, we introduce a new variant of the BFGS method designed to perform well when gradient measurements are corrupted by noise. We show that by treating the secant conditi…
How To Catch A Lion In The Desert -- On The Solution Of The Coverage Directed Generation (CDG) Problem
Raviv Gal, Eldad Haber, Brian Irwin +2
The testing and verification of a complex hardware or software system, such as modern integrated circuits (ICs) found in everything from smartphones to servers, can be a difficult…