Pith. sign in

REVIEW

On Matching Pursuit and Coordinate Descent

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 1803.09539 v7 pith:FUHV3WKE submitted 2018-03-26 stat.ML cs.LGmath.OC

classification stat.MLcs.LGmath.OC
keywords coordinatedescentmatchingpursuitobjectivesaffinealgorithmsanalysis
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

Two popular examples of first-order optimization methods over linear spaces are coordinate descent and matching pursuit algorithms, with their randomized variants. While the former targets the optimization by moving along coordinates, the latter considers a generalized notion of directions. Exploiting the connection between the two algorithms, we present a unified analysis of both, providing affine invariant sublinear $\mathcal{O}(1/t)$ rates on smooth objectives and linear convergence on strongly convex objectives. As a byproduct of our affine invariant analysis of matching pursuit, our rates for steepest coordinate descent are the tightest known. Furthermore, we show the first accelerated convergence rate $\mathcal{O}(1/t^2)$ for matching pursuit and steepest coordinate descent on convex objectives.

Discussion (0). Sign in to comment.

Pith tools