OMLSP 2013 - 2013 Symposium on: Optimization in Machine Learning and Signal Processing

This workshop seeks to explore the fertile intersection of signal processing, machine learning and large-scale optimization. Many recent and fundamental advances in drawing inference from data have involved formulating statistical objectives - both inferring a model, and predicting thereof - as optimization problems.
A key factor complicating matters is that modern signal processing applications often demand solving such optimization problems at very large scales. It thus becomes important to leverage any structure present in the statistical estimation problem, both classical structures such as strict/strong convexity, and smoothness, but also other statistical-estimation-specific structures such as sparsity, graphical model structure, low-rank structure and so on.
What are the operational and fundamental relationships between computation and statistical efficiency? Which facets of optimization methods are inherently parallelizable (e.g. message-passing algorithms)? What are the limits and bottlenecks faced by the state of the art optimization methods when faced with large-scale problems? It is the aim of this workshop to answer questions in this vein by bringing together researchers from different communities --- signal processing, machine learning, statistics and mathematical programming --- and identify common intuitions underlying successful methods.
Submissions of at most 4 pages in two-column IEEE format are welcome on topics including:
Models and estimation
Sparsity, Low-rank and other methods in high-dimensional statistics
Large-scale convex optimization: algorithms and applications
Graphical models: inference, structure learning etc.
Optimization for clustering, classification, regression etc.
Non-convex and iterative methods