A Hybrid Algorithm for Convex Semidefinite Optimization

About

We present a hybrid algorithm for optimizing a convex, smooth function over the cone of positive semidefinite matrices. Our algorithm converges to the global optimal solution and can be used to solve general large-scale semidefinite programs and hence can be readily applied to a variety of machine learning problems. We show experimental results on three machine learning problems (matrix completion, metric learning, and sparse PCA) . Our approach outperforms state-of-the-art algorithms.

Soeren Laue · 2012

Related benchmarks

Task	Dataset	Result
Matrix completion	MovieLens 1M (test)	RMSE0.874	40
Matrix completion	MovieLens-100K (test)	RMSE0.933	24
Sparse PCA	Colon cancer data	F Value-1.8675	14
Metric Learning	Synthetic Data	F-Value0.000337	6
Clustering	SONAR	Clustering Accuracy91.59	4
Clustering	Wine	Clustering Accuracy84.14	4
Clustering	chessboard	Clustering Accuracy91.28	4
Clustering	dbl-spiral	Clustering Accuracy99.7	4
Clustering	Glass	Clustering Accuracy83.7	4
Clustering	Iris	Clustering Accuracy98.69	4