Parallel Machine Learning Approaches for Reverse Engineering Genome-scale Networks Srinivas Aluru, Georgia Institute of Technology, USA

Abstract: Reverse engineering whole-genome networks from large-scale gene expression measurements and analyzing them to extract biologically valid hypotheses are important challenges in systems biology. While simpler models easily scale to large number of genes and gene expression datasets, more accurate models are compute intensive limiting their scale of applicability. In this talk, I will present our research on the development of parallel mutual information and Bayesian network based structure learning methods to eliminate such bottlenecks and facilitate learning of genome-scale networks. Such networks can be used as a guide to predicting gene function and extracting context-specific subnetworks.

Bio: Srinivas Aluru is a professor in the School of Computational Science and Engineering at Georgia Institute of Technology. He co-directs the Georgia Tech Interdisciplinary Research Institute in Data Engineering and Science (IDEaS), and co-leads the NSF South Big Data Regional Innovation Hub which serves 16 Southern States in the U.S. and Washington D.C. Aluru conducts research in high performance computing, bioinformatics and systems biology, combinatorial scientific computing, and applied algorithms. He is currently serving as the Chair of the ACM Special Interest Group on Bioinformatics, Computational Biology and Biomedical Informatics (SIGBIO). He is a Fellow of the AAAS and IEEE.

Graphs and Sparse Matrices: There and Back Again John Gilbert, UC Santa Barbara, USA

Abstract: The mathematical connections between graph theory and linear algebra are intimate and well known. The computational links between the two fields are also deep, extending from basic data structures to fundamental decompositions to the design of efficient algorithms.
During the first 50 years of this computational relationship, graphs served numerical linear algebra by enabling efficient sparse matrix computation. Recently, matrix computation has been returning the favor. I will talk about the past and present of the relationship in both directions, and speculate a bit on its future.

Bio: John R. Gilbert directs the Combinatorial Scientific Computing Laboratory at the University of California at Santa Barbara, where he is Professor of Computer Science. He has done fundamental work in algorithms and software for combinatorial and numerical problems and for sparse matrix computation, including Matlabâ€™s original sparse matrix capabilities and the SuperLU solver library. His current research applies linear algebraic methods to large-scale graph and network analytics.
Professor Gilbert received his PhD from Stanford University in 1981. He has been on the Computer Science faculty at Cornell, a visiting faculty member at MIT, and a Principal Scientist at Xerox PARC, where he directed research in algorithms, distributed sensing and control, and robotics. He has chaired the ACM Special Interest Group on Numerical Mathematics and the SIAM Activity Group on Supercomputing, and is a Fellow of the Society for Industrial and Applied Mathematics.

The Revolution in Graph Theoretic Optimization Gary Miller, Carnegie Mellon University, USA

Abstract: Graph theoretic optimization problems that have been dormant for
fifty years are now seeing new and exciting algorithms. These advances
have been made possible by Spectral Graph Theory, the interplay
between linear algebra and combinatorial graph theory. One application
of this interplay is a nearly linear time solver for Symmetric
Diagonally Dominant systems (SDD). This seemingly restrictive class
of linear systems has received substantial interest in the last fifteen
years, and there is an ever growing list of
problems that are amenable to SDD solvers, including
image segmentation, image denoising, finding solutions to elliptic
equations, computing maximum flow in a graph, and graph sparsification.
All these examples can be viewed as special cases of
convex optimization problems on graphs.

Possibly the best known graph theoretic optimization problem is
computing the maximum flow in a graph. Again, using spectral graph
theory, the maximum flow problem in undirected graphs can now be
approximately solved in almost linear time. Even the harder problem
of directed graph flow has seen improvements in the last year.
I claim that this is only the beginning of a new era in efficient
algorithm design.
In this talk I will mostly focus the ideas and methods used in these
new solvers. It is interesting that they are a hybrid between classic
numerical methods and new ideas and algorithms from graph theory, but from
a spectral prospective.

Bio: Gary Miller is a professor of Computer Science at Carnegie Mellon University.