Date of Award

12-1-2012

Document Type

Campus Access Thesis

Degree Name

Master of Science (MS)

Department

Computer Science

First Advisor

Wei Ding

Second Advisor

Dan A. Simovici

Third Advisor

Marc Pomplun

Abstract

The least squares problem is one of the most important regression problems in statistics and machine learning. In this paper, we present an Averaging Projection Stochastic Gradient Descent (APSGD) algorithm to solve the large-scale least squares problem. APSGD improves the Stochastic Gradient Descent (SGD) by using the constraint that the linear regression line passes through the mean point of all the data points. It results in the best regret bound O(logT), and fastest convergence speed among all first order approaches. Empirical studies confirm the effectiveness of APSGD by comparing it with the state-of-art methods.

Comments

Free and open access to this Campus Access Thesis is made available to the UMass Boston community by ScholarWorks at UMass Boston. Those not on campus and those without a Healey Library (UMass Boston) barcode may gain access to this thesis through resources like Proquest Dissertations & Theses Global. If you have a Healey Library barcode and would like to download this work from off-campus, click on the "Off-Campus UMass Boston Users" link above.

Share

COinS