We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Machine Learning

Title: Identifying global optimality for dictionary learning

Abstract: Learning new representations of input observations in machine learning is often tackled using a factorization of the data. For many such problems, including sparse coding and matrix completion, learning these factorizations can be difficult, in terms of efficiency and to guarantee that the solution is a global minimum. Recently, a general class of objectives have been introduced-which we term induced dictionary learning models (DLMs)-that have an induced convex form that enables global optimization. Though attractive theoretically, this induced form is impractical, particularly for large or growing datasets. In this work, we investigate the use of practical alternating minimization algorithms for induced DLMs, that ensure convergence to global optima. We characterize the stationary points of these models, and, using these insights, highlight practical choices for the objectives. We then provide theoretical and empirical evidence that alternating minimization, from a random initialization, converges to global minima for a large subclass of induced DLMs. In particular, we take advantage of the existence of the (potentially unknown) convex induced form, to identify when stationary points are global minima for the dictionary learning objective. We then provide an empirical investigation into practical optimization choices for using alternating minimization for induced DLMs, for both batch and stochastic gradient descent.
Comments: Updates to previous version include a small modification to Proposition 2, to only use normed regularizers, and a modification to the main theorem (previously Theorem 13) to focus on the overcomplete, full rank setting and to better characterize non-differentiable induced regularizers. The theory has been significantly modified since version 1
Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
Cite as: arXiv:1604.04942 [stat.ML]
  (or arXiv:1604.04942v4 [stat.ML] for this version)

Submission history

From: Martha White [view email]
[v1] Sun, 17 Apr 2016 23:46:04 GMT (2121kb,D)
[v2] Thu, 6 Oct 2016 01:13:04 GMT (2261kb,D)
[v3] Sun, 19 Feb 2017 16:10:17 GMT (3412kb,D)
[v4] Mon, 7 Aug 2017 02:04:52 GMT (2756kb,D)

Link back to: arXiv, form interface, contact.