From pondabarnes at tti-c.org Wed Feb 7 15:33:01 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Wed Feb 7 15:33:13 2007 Subject: [TTIC Colloquium] Guest speaker announcement Message-ID: <200702072133.l17LXAsD017377@nagoya.uchicago.edu> TTI-C Guest Speaker Presented by: Toyota Technological Institute Speaker: Mehmet Koyuturk Speaker's home page: http://www.cs.purdue.edu/people/faculty/koyuturk/ Date: Thursday, February 8, 2007 Location: TTI Conference room, 2nd floor- Press Building Time: 10:00am Title: Comparative Analysis of Molecular Interaction Network Abstract: Emergence of high-throughput experiments and resulting databases capture relationships and interactions between biomolecules. These interactions enable modeling and analysis of a cell from a systems perspective - generally using network models. In this talk, we focus on development of computational tools and statistical models for comparative analysis of molecular interaction networks. These tools target understanding of functional modularity in the cell by extracting novel information from massive amounts of interaction data, through integration of cellular organization, functional hierarchy, and evolutionary conservation. We first discuss the problem of identifying conserved sub-networks in a collection of interaction networks belonging to diverse species. The main algorithmic challenges here stem from the NP-hard sub graph isomorphism problem that underlies frequent sub graph discovery. Three decades of research into theoretical aspects of this problem has highlighted the futility of syntactic approaches, thus motivating use of semantic information. Using a biologically motivated homolog contraction technique for relating proteins across species, we render this problem tractable. We experimentally show that the proposed method can be used as a pruning heuristic that accelerates existing techniques significantly, as well as a standalone tool that conveys significant biological insights at near-interactive rates. With a view to understanding the conservation and divergence of modular substructures, we also develop network alignment techniques, grounded in theoretical models of network evolution. Through graph-theoretic modeling of evolutionary events in terms of matches, mismatches, and duplications, we reduce the alignment problem to a graph optimization problem and develop heuristics to solve this problem efficiently. In order to assess the statistical significance of the patterns identified by our algorithms, we probabilistically analyze the distribution of highly connected and conserved sub graphs in random graphs. Our methods and algorithms are implemented on various platforms and tested extensively on a comprehensive collection of molecular interaction data, illustrating their effectiveness in terms of providing novel biological insights as well as computational efficiency. This is joint work with Yohan Kim, Shankar Subramaniam (University of California, San Diego), Wojciech Szpankowski, and Ananth Grama (Purdue University) and is supported by the National Institutes of Health. If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future talks and events, please go to the TTI-C events calendar at http://ttic.uchicago.edu/cal/month.php -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20070207/43c986af/attachment-0001.htm From pondabarnes at tti-c.org Mon Feb 12 10:12:09 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Mon Feb 12 10:12:31 2007 Subject: [TTIC Colloquium] Guest speaker announcement Message-ID: <200702121612.l1CGCTsD022173@nagoya.uchicago.edu> TTI-C Guest Speaker Presented by: Toyota Technological Institute Speaker: Vitaly Feldman Speaker's home page: http://www.eecs.harvard.edu/~vitaly/ Date: Tuesday, February 13, 2007 Location: TTI-C Conference room, 2nd floor-Press Building Time: 10:00am Title: Approaches to the DNF Learning Problem Abstract: Learning of Disjunctive Normal Form (DNF) formulas is one of the most fundamental problems in computational learning theory. Informally, in this problem the goal of a learning algorithm is to accurately predict an unknown two-level (OR of ANDs) Boolean formula given examples labeled by this formula. In the first part of this talk, I will survey some of the fundamental results concerning efficient learnability of DNF in Valiant's PAC model as well as the progress that was made in recent years. Learning of parities (XOR functions) with random noise with respect to the uniform distribution is another famous open problem in learning theory and is equivalent to decoding of random linear codes - a major open problem in coding theory. In the second part of this talk, I will describe a recent result that shows a surprising connection between learning of parity functions with random noise and learning of DNF from random and uniform examples. I will also show other applications of the key technical component of this result. The talk is largely self-contained. If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future talks and events, please go to the TTI-C events calendar at http://ttic.uchicago.edu/cal/month.php -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20070212/85651d17/attachment.htm From pondabarnes at tti-c.org Mon Feb 12 10:28:52 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Mon Feb 12 10:29:14 2007 Subject: [TTIC Colloquium] FW: Guest speaker announcement Message-ID: <200702121629.l1CGTDsD022377@nagoya.uchicago.edu> TTI-C Guest Speaker Presented by: Toyota Technological Institute Speaker: Vitaly Feldman Speaker's home page: http://www.eecs.harvard.edu/~vitaly/ Date: Tuesday, February 13, 2007 Location: TTI-C Conference room, 2nd floor-Press Building Time: 10:00am Title: Approaches to the DNF Learning Problem Abstract: Learning of Disjunctive Normal Form (DNF) formulas is one of the most fundamental problems in computational learning theory. Informally, in this problem the goal of a learning algorithm is to accurately predict an unknown two-level (OR of ANDs) Boolean formula given examples labeled by this formula. In the first part of this talk, I will survey some of the fundamental results concerning efficient learnability of DNF in Valiant's PAC model as well as the progress that was made in recent years. Learning of parities (XOR functions) with random noise with respect to the uniform distribution is another famous open problem in learning theory and is equivalent to decoding of random linear codes - a major open problem in coding theory. In the second part of this talk, I will describe a recent result that shows a surprising connection between learning of parity functions with random noise and learning of DNF from random and uniform examples. I will also show other applications of the key technical component of this result. The talk is largely self-contained. If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future talks and events, please go to the TTI-C events calendar at http://ttic.uchicago.edu/cal/month.php -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20070212/a25f1703/attachment.htm From pondabarnes at tti-c.org Thu Feb 15 09:12:44 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Thu Feb 15 09:13:54 2007 Subject: [TTIC Colloquium] Guest speaker announcement Message-ID: <200702151513.l1FFDpIq011321@nagoya.uchicago.edu> TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Julia Chuzhoy Speaker's home page: http://www.math.ias.edu/~cjulia/ Date: Thursday, February 15, 2007 Location: TTI-C Conference room, 2nd floor Press Building Time: 10:00am Title: Cuts and Flows in Directed Graphs Abstract: Cuts and flows are among the most basic graph theoretic notions. Applications that require solving graph cut or flow problems arise in almost every area of computer science. The study of the connection between flows and cuts dates back to the late fifties when Ford and Fulkerson proved that in the single-commodity environment, minimum cut equals maximum flow in any graph. A natural generalization of this result would be establishing the relationship between flows and cuts in the presence of multiple commodities. This relationship is usually expressed via the notion of flow-cut gap: the maximum ratio, achievable for any graph, between the maximum multi-commodity flow and the corresponding cut value, called minimum multicut. Flow-cut gaps have been extensively studied for more than five decades now, and they are widely used in the design and the analysis of algorithms. One of the major breakthroughs in this area is a complete understanding of the flow-cut gap in undirected graphs, which was proved logarithmic. In spite of this success, the flow-cut gaps have remained poorly understood in directed graphs. In particular, it has remained an open question whether the flow-cut gap in directed graphs is also logarithmic. In this talk, we will answer this question in the negative by showing that, in sharp contrast to the undirected case, the flow-cut gap in directed graphs is polynomial. If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future talks and events, please go to the TTI-C events calendar at http://ttic.uchicago.edu/cal/month.php -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20070215/cdec5a98/attachment-0001.htm From pondabarnes at tti-c.org Tue Feb 20 14:52:10 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Tue Feb 20 14:56:29 2007 Subject: [TTIC Colloquium] Guest speaker announcement Message-ID: <200702202056.l1KKuRIq016031@nagoya.uchicago.edu> TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Joel Tropp Speaker's home page: http://www-personal.umich.edu/~jtropp Date: Monday, 26, 2007 Location: TTI-C Conference room, 2nd floor-Press building Time: 10:00 Title: Sparse Solutions to Underdetermined Linear Systems Abstract: Title: "Sparse solutions to underdetermined linear systems" Abstract: A fundamental problem in applied mathematics, statistics, and electrical engineering is to solve underdetermined systems of linear equations. Basic linear algebra seems to forbid this possibility. But a recent strand of research has established that certain underdetermined systems can be solved robustly with efficient algorithms, provided that the solution is sparse (i.e., has many zero components). This talk provides an overview of these sparse representation problems, and it describes the basic algorithmic approaches. Then it details situations where the algorithms are guaranteed to succeed. In particular, the talk introduces some new work on the case where the matrix is deterministic and the sparsity pattern is random. It also covers some results for the case where the matrix is random, which is the situation in Compressed Sensing applications. If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@ tti-c.org. For future talks and events, please go to the TTI-C events calendar at http://ttic.uchicago.edu/cal/month.php -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20070220/89319086/attachment.htm From pondabarnes at tti-c.org Mon Feb 26 09:54:39 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Mon Feb 26 10:02:46 2007 Subject: [TTIC Colloquium] Guest Speaker Message-ID: <200702261602.l1QG2iIq016549@nagoya.uchicago.edu> TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Joel Tropp Speaker's home page: http://www-personal.umich.edu/~jtropp/ Date: Monday, February 26, 2007 Location: TTI-C Conference room Time: 10:00am Title: "Sparse solutions to underdetermined linear systems" Abstract: A fundamental problem in applied mathematics, statistics, and electrical engineering is to solve underdetermined systems of linear equations. Basic linear algebra seems to forbid this possibility. But a recent strand of research has established that certain underdetermined systems can be solved robustly with efficient algorithms, provided that the solution is sparse (i.e., has many zero components). This talk provides an overview of these sparse representation problems, and it describes the basic algorithmic approaches. Then it details situations where the algorithms are guaranteed to succeed. In particular, the talk introduces some new work on the case where the matrix is deterministic and the sparsity pattern is random. It also covers some results for the case where the matrix is random, which is the situation in Compressed Sensing applications. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20070226/7d037ec2/attachment.htm From pondabarnes at tti-c.org Mon Feb 26 10:23:51 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Mon Feb 26 10:31:54 2007 Subject: [TTIC Colloquium] Guest speaker announcement Message-ID: <200702261631.l1QGVrIq016596@nagoya.uchicago.edu> TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Hongwei Wu Speaker's homepage: http://csbl.bmb.uga.edu/~hongweiw/ Date: Tuesday, February 27, 2007 Location: TTI-Conference room Time: 10:00 Title: Multi-level Functional Classification of Genes and Prediction of Functional Modules for Bacterial Genomes Abstract: Since late eighties and early nineties when the Human Genome Project was initiated, a large volume of genomic data of different organisms have been made available thanks to the world-wide sequencing efforts; and, with the development of high-throughout experiment technologies, a lot of measurements about functional and structural properties of biological molecules have also been made available. In this post-genome era, the focuses of the field of computational biology and bioinformatics are to use computer-based methods to analyze and interpret these data. Biological functions of genes can be described from two perspectives. One perspective is to describe the activities of genes and their products at the molecular level; and the other perspective is to describe the roles of genes and their products in the biological processes that they participate in. Accordingly, there are generally two kinds of methods to predict biological functions of newly sequenced genes. One is to identify the genes in those well-investigated genomes that are similar to the unknown genes; and the other is to identify the genes that are functionally related to the unknown genes. This talk will focus on my work on the multi-level functional classification of genes and the prediction of functional modules for bacterial genomes, which belong to the two different kinds of methods for the prediction of gene functions, respectively. Our studies on the multi-level functional classification of genes can not only be used to provide functional annotations of newly sequenced gene from multiple resolution levels, but can also be used to reveal evolutionary trace of genes and genomes. Whereas, our studies on the prediction of functional modules can not only be used to reveal the functional relatedness between genes, but also represent a key step towards deciphering biological networks/pathways in a systematic way. If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future TTI-C talks and events, please go to http://ttic.uchicago.edu/cal/month.php -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20070226/3d38e2ab/attachment-0001.htm From pondabarnes at tti-c.org Tue Feb 27 09:53:48 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Tue Feb 27 10:02:22 2007 Subject: [TTIC Colloquium] FW: Guest speaker announcement Message-ID: <200702271602.l1RG2JIq019975@nagoya.uchicago.edu> Reminder!! TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Hongwei Wu Speaker's homepage: http://csbl.bmb.uga.edu/~hongweiw/ Date: Tuesday, February 27, 2007 Location: TTI-Conference room Time: 10:00 Title: Multi-level Functional Classification of Genes and Prediction of Functional Modules for Bacterial Genomes Abstract: Since late eighties and early nineties when the Human Genome Project was initiated, a large volume of genomic data of different organisms have been made available thanks to the world-wide sequencing efforts; and, with the development of high-throughout experiment technologies, a lot of measurements about functional and structural properties of biological molecules have also been made available. In this post-genome era, the focuses of the field of computational biology and bioinformatics are to use computer-based methods to analyze and interpret these data. Biological functions of genes can be described from two perspectives. One perspective is to describe the activities of genes and their products at the molecular level; and the other perspective is to describe the roles of genes and their products in the biological processes that they participate in. Accordingly, there are generally two kinds of methods to predict biological functions of newly sequenced genes. One is to identify the genes in those well-investigated genomes that are similar to the unknown genes; and the other is to identify the genes that are functionally related to the unknown genes. This talk will focus on my work on the multi-level functional classification of genes and the prediction of functional modules for bacterial genomes, which belong to the two different kinds of methods for the prediction of gene functions, respectively. Our studies on the multi-level functional classification of genes can not only be used to provide functional annotations of newly sequenced gene from multiple resolution levels, but can also be used to reveal evolutionary trace of genes and genomes. Whereas, our studies on the prediction of functional modules can not only be used to reveal the functional relatedness between genes, but also represent a key step towards deciphering biological networks/pathways in a systematic way. If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future TTI-C talks and events, please go to http://ttic.uchicago.edu/cal/month.php -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20070227/9c240a5b/attachment.htm