Quasi-Newton Optimization Methods For Deep Learning Applications

no code implementations • 4 Sep 2019 • Jacob Rafati, Roummel F. Marcia

Quasi-Newton methods, like SGD, require only first-order gradient information, but they can result in superlinear convergence, which makes them attractive alternatives to SGD.

Image Classification reinforcement-learning +2

Deep Reinforcement Learning via L-BFGS Optimization

no code implementations • 6 Nov 2018 • Jacob Rafati, Roummel F. Marcia

Deep Reinforcement Learning algorithms require solving a nonconvex and nonlinear unconstrained optimization problem.

Atari Games Q-Learning +3

Trust-Region Algorithms for Training Responses: Machine Learning Methods Using Indefinite Hessian Approximations

no code implementations • 1 Jul 2018 • Jennifer B. Erway, Joshua Griffin, Roummel F. Marcia, Riadh Omheni

Machine learning (ML) problems are often posed as highly nonlinear and nonconvex unconstrained optimization problems.

BIG-bench Machine Learning

Compressive Coded Aperture Keyed Exposure Imaging with Optical Flow Reconstruction

no code implementations • 26 Jun 2013 • Zachary T. Harmany, Roummel F. Marcia, Rebecca M. Willett

This paper describes a coded aperture and keyed exposure approach to compressive video measurement which admits a small physical platform, high photon efficiency, high temporal resolution, and fast reconstruction algorithms.

Optical Flow Estimation Video Reconstruction