- talks - lists.cs.princeton.edu

Colloquium Speaker Jianxiong Xiao, Wed March 13, 4:30pm
by Nicole E. Wagenblast 13 Mar '13

13 Mar '13

A 2D + 3D Rich Data Approach to Scene Understanding Jianxiong Xiao , Massachusetts Institute of Technology Wednesday, March 13, 2013, 4:30pm Computer Science 105 On your one-minute walk from the coffee machine to your desk each morning, you pass by dozens of scenes -- a kitchen, an elevator, your office -- and you effortlessly recognize them and perceive their 3D structure. But this one-minute scene-understanding problem has been an open challenge in computer vision for decades. Recently, researchers have come to realize that big data is critical for building scene-understanding systems that can recognize the semantics and reconstruct the 3D structure. In this talk, I will share my experience in leveraging big data for scene understanding, shifting the paradigm from 2D view-based categorization to 3D place-centric representations. To push the traditional 2D representation to the limit, we built the Scene Understanding (SUN) Database, a large collection of images that exhaustively spans all scene categories. However, the lack of a "rich" representation still significantly limits the traditional recognition pipeline. While an image is a 2D array, the world is 3D and our eyes see it from a viewpoint, but this is not traditionally modeled. This paradigm shift toward rich representation also opens up new challenges that require a new kind of big data -- data with extra descriptions, namely rich data. Specifically, we focus on a highly valuable kind of rich data -- multiple viewpoints in 3D -- and we build the SUN3D database to obtain an integrated "place-centric" representation of scenes. This novel representation with rich data opens up exciting new opportunities for integrating scene recognition over space and for obtaining a scene-level reconstruction of large environments. It also has many applications such as organizing big visual data to provide photo-realistic indoor 3D maps. Finally, I will discuss some open challenges and my future plans for rich data and representation. Jianxiong Xiao is a Ph.D. candidate in the Computer Science and Artificial Intelligence Laboratory (CSAIL) at Massachusetts Institute of Technology (MIT). Before that, he received a B.Eng. and a M.Phil. from the Hong Kong University of Science and Technology. His research interests are in computer vision, with a focus on scene understanding. His work has received the Best Student Paper Award at the European Conference on Computer Vision (ECCV) in 2012, and has appeared in popular press. Jianxiong was awarded the Google U.S./Canada Ph.D. Fellowship in Computer Vision in 2012 and MIT CSW Best Research Award in 2011. More information can be found on his website: http://mit.edu/jxiao .

1 3

Colloquium Speaker Anshul Kundaje Tues March 12, 4:30pm
by Nicole E. Wagenblast 12 Mar '13

12 Mar '13

Unraveling the heterogeneity and dynamics of regulatory elements in the human genome Anshul Kundaje , Massachusetts Institute of Technology Tuesday, March 12, 2013, 4:30pm Computer Science 105 In 2003, the Human Genome Project marked a major scientific milestone by releasing the first consensus DNA sequence of the human genome. The ENCODE Project (Encyclopedia of DNA elements) was launched to pick up where the Human Genome Project left off, with the ambitious goal of systematically deciphering the potential function of every base (letter) in the genome. ENCODE has generated the largest collection of functional genomic data in humans to date, measuring the cellular activity of thousands of cellular moieties in a variety of normal and diseased cellular contexts. In this talk, I will describe novel computational and machine learning approaches that I developed for integrative analysis of massive compendia of diverse biological data such as ENCODE to unravel the functional heterogeneity and variation of regulatory elements in the human genome and their implications in human disease. I will begin with a gentle introduction to the diversity and scale of ENCODE data and a brief overview of robust, statistical methods that we developed for automated detection of DNA binding sites of hundreds of regulatory proteins from noisy, experimental data. Regulatory proteins can perform multiple functions by interacting with and co-binding DNA with different combinations of other regulatory proteins. I developed a novel discriminative machine learning formulation based on regularized Rule-based ensembles that was able to sort through the combinatorial complexity of possible regulatory interactions and learn statistically significant item-sets of co-binding events at an unprecedented level of detail. I found extensive evidence that regulatory proteins could switch partners at different sets of genomic domains within a single cell-type and across different cell-types affecting structural and chemical properties of DNA and regulating different functional categories of target genes. Using regulatory elements discovered from ENCODE data, we were also able to provide putative functional interpretations for up to 81% of all publicly available sequence variants (mutations) identified in large-scale disease studies and generate new hypotheses by integrating multiple sources of data. Finally, I will present a brief overview of my recent efforts on using multivariate Hidden Markov models to analyze the dynamics of various chemical modifications to DNA across three key axes of variation - across multiple species, across different cell-types in a single species (human), and across multiple human individuals for the same cell-type. Our results indicate a remarkable universality of chemical modifications defining hidden regulatory states across the animal kingdom with dramatic differences in the variation and functional impact of these regulatory elements between cell-types and individuals. Together, these efforts take us one step closer to learning comprehensive models of gene regulation in humans in order to improve our system-level understanding of cellular processes and complex diseases.

1 3

Colloquium speaker Aditya Parameswaran, Mon March 12, 4:30pm
by Nicole E. Wagenblast 11 Mar '13

11 Mar '13

Human-Powered Data Management Aditya Parameswaran , Stanford University Monday, March 12, 2013, 4:30pm Computer Science 105 Fully automated algorithms are inadequate for a number of data analysis tasks, especially those involving images, video, or text. Thus, there is a need to combine “human computation” (or crowdsourcing), together with traditional computation, in order to improve the process of understanding and analyzing data. My thesis addresses several topics in the general area of human-powered data management. I design algorithms and systems for combining human and traditional computation for: (a) data processing, e.g., using humans to help sort, cluster, or clean data; (b) data extraction, e.g., having humans help create structured data from information in unstructured web pages; and (c) data gathering, i.e., asking humans to provide data that they know about or can locate, but that would be difficult to gather automatically. My focus in all of these areas is to find solutions that expend as few resources as possible (e.g., time waiting, human effort, or money spent), while still providing high quality results. In this talk, I will first present a broad perspective of our research on human-powered data management, and I will describe some systems and applications that have motivated our research. I will then present details of one of the problems we have addressed: filtering large data sets with the aid of humans. Finally I will argue that human-powered data management is an area in its infancy, by describing a number of open problems I intend to address in my future research program. Aditya Parameswaran is a Ph.D. student in the InfoLab at Stanford University, advised by Prof. Hector Garcia-Molina. He is broadly interested in data management, with research results in human computation, information extraction, and recommendation systems. Aditya is a recipient of the Key Scientific Challenges Award from Yahoo! Research (2010), two best-of-conference citations (VLDB 2010 and KDD 2012), and the Terry Groswith graduate fellowship at Stanford University.

1 3

S Sachdeva preFPO
by Melissa M. Lawson 07 Mar '13

07 Mar '13

Sushant Sachdeva will present his preFPO on Tuesday March 12 at 11AM in Room 302 (note room!). The members of his committee are: Sanjeev Arora, advisor; Moses Charikar and Nisheeth Vishnoi (MSR India) readers; Beranrd Chazelle and Mark Braverman, nonreaders. Everyone is invited to attend his talk. His abstract follows below. ============================================== New Results in Graph Partitioning and Optimal Inapproximability sans UGC This thesis consists of two parts. The first part contains improved algorithms for graph partitioning problems, both for well-studied classical problems (Balanced Separator) and for more practically motivated problems (Identifying Communities). The second part contains new results towards achieving optimal inapproximability results without assuming the Unique Games Conjecture. A. Graph Partitioning We give the first near-linear time algorithm for Balanced Separator with the best possible guarantee for spectral algorithms. The algorithm combines SDP rounding techniques with continuous time random walks on the graph. We show how to approximate these walks, or equivalently, the matrix exponential e^{-tL}, where L is the graph Laplacian, in near-linear time, by combining techniques from approximation theory and numerical linear algebra. We give a novel approach to the problem of identifying communities in a graph representing a social network. We model the network as being described by overlapping communities with certain local structure; and under these assumptions, we provably and efficiently recover the underlying communities. B. Optimal Inapproximability without assuming UGC We study the inapproximability of some classic scheduling problems viz. Concurrent Open Shop and the Assembly Line problem, and prove an optimal 2-inapproximability for both problems assuming P != NP. Previously, the best NP-hardness factors known were 6/5 and 1+eps respectively. We also study the inapproximability of Hypergraph Vertex Cover on k-uniform k-partite hypergraphs, and prove a nearly optimal NP-hardness of approximating it within k/2-1, where a k/2-approximation is known due to Lovász. The results in this talk are based on joint works with Sanjeev Arora, Rong Ge, Lorenzo Orecchia, Rishi Saket, Grant Schoenebeck, Madhur Tulsiani, and Nisheeth K. Vishnoi.

1 0

R Ge preFPO
by Melissa M. Lawson 06 Mar '13

06 Mar '13

Rong Ge will present his preFPO on Monday March 11 at Noon in Room 401 (note room!) The members of his committee are: Sanjeev Arora, advisor; Moses Charikar and Sham Kakade (MSR New England), readers; Mark Braverman and Zeev Dvir, nonreaders. Everyone is invited to attend his talk. His abstract follows below. --------------------------- Title: Provable Algorithms for Machine Learning Many tasks in machine learning are computationally intractable. Nevertheless, researchers have developed heuristic algorithms that successfully solve these problems in practice. This phenomena can be explained using average case complexity. However, in many situations the details supporting this explanation are still missing: what properties of the input make it possible to be solved efficiently? What is the chance that real-life inputs satisfy these properties and how robust is the algorithm if they don't? In this talk I use topic modeling as an example to show our attempts of answering these problems. The first part of the talk gives a new algorithm for learning topic models with provable guarantees. This algorithm relies on a reasonable simplifying assumption, and runs 20 times faster than existing software like mallet with comparable output quality. The second part of the talk gives a new analysis for method of moments algorithms (AHK12, AFHKL12) using tensor decomposition. The new algorithm has a better perturbation analysis and can therefore tolerate more modeling error.

1 0

Colloquium Speaker Floraine Berthouzoz Mon March 4, 4:30pm
by Nicole E. Wagenblast 04 Mar '13

04 Mar '13

Procedural Instructions for People and Machines Floraine Berthouzoz , University of California, Berkeley Monday, March 4, 2013, 4:30pm Computer Science, 105 Procedural tasks such as following a recipe or editing an image are very common. They require a person to execute a sequence of operations (e.g. chop the onions, or sharpen the image) and people commonly use step-by-step tutorials to learn these tasks. We focus on the domain of photo manipulations and have developed tools and techniques to help people learn, compare and automate photo manipulation procedures. We present a demonstration-based system for automatically generating succinct step-by-step visual tutorials of photo manipulations. Our tutorials are quick to generate. Moreover, users are faster and make fewer errors with our tutorials compared to book and video tutorials. We also demonstrate a new interface that allows learners to browse, analyze and compare large collections (i.e. thousands) of photo manipulation tutorials based on their command-level structure. Finally, we present a framework for generating content-adaptive macros (programs) that can transfer complex photo manipulation procedures to new target images. Together these tools allow people to learn, compare and automate procedural knowledge. Floraine Berthouzoz is a Ph.D. candidate in the Computer Science Department at the University of California, Berkeley. Her research lies at the intersection of computer graphics and human-computer interactions, and aims at building tools that ultimately make it faster and easier for people to create high quality visual content. She has also developed a number of techniques that improve the way people learn procedural instructions. She is the recipient of a 2009 ETH Medal of Honor and a 2010 Google Anita Borg Memorial Scholarship. During her graduate studies, Floraine worked as a visiting researcher at the University of Tokyo, the Hebrew University, Columbia University and Adobe Systems. Floraine also co-founded CS KickStart, an outreach program for incoming undergraduate women. Since 2011, this program has significantly increased the number of undergraduate women pursuing computer science degrees at Berkeley.

1 1

A Bell preFPO
by Melissa M. Lawson 28 Feb '13

28 Feb '13

Ana Bell will present her preFPO on Friday March 1 (this Friday) at 10AM in room 302 (note room). The members of her committee are: Olga Troyanskaya, advisor; Mona Singh and Andrea LaPaugh, readers; Kai Li and Tom Funkhouser, nonreaders. Everyone is invited to attend her talk. Her abstract follows below. ----- Original Message ----- Title and Abstract: Integrating Genomic Data to Build Mechanistic Networks for Genes and Small Molecules The recent increase in interest in mining biological data has spurred the aggregation of genomic data from various model organisms into numerous databases. We want to integrate the data available for one of the more well studied organisms, saccharomyces cerevisiae (yeast), and to build mechanistic networks that present interactions between proteins and compounds (small molecules) so that we can accurately predict drug targets. To that end, we apply machine learning algorithms to the data to create the interaction network; a graph where nodes represent proteins or molecules and edges between nodes represent the probability that two nodes interact. Our two step integration process, where we first predict protein-protein interaction networks for various interaction types and then use these networks to predict protein-compound interaction networks, will provide detailed insight into how pathway level knowledge can be leveraged to predict small molecule level interactions. Recent years have also seen an explosion in plant genomics, as the difficulties inherent in sequencing and functionally analyzing these biologically and economically significant organisms have been overcome. Arabidopsis thaliana, a versatile model organism, represents an opportunity to evaluate the predictive power of biological network inference for plant functional genomics. We provide a compendium of functional relationship networks for A. thaliana leveraging data integration based on over 60 microarray, physical and genetic interaction, and literature curation datasets. These include tissue, biological process, and development stage specific networks, each predicting relationships specific to an individual biological context. These biological networks enable the rapid investigation of uncharacterized genes in specific tissues and developmental stages of interest and summarize a very large collection of A. thaliana data for biological examination. We found validation in the literature for many of our predicted networks, including those involved in disease resistance, root hair patterning, and auxin homeostasis.

1 1

Colloquium Speaker Ankur Moitra Wed Feb 27- 4:30pm
by Nicole E. Wagenblast 27 Feb '13

27 Feb '13

New Algorithms for Nonnegative Matrix Factorization and Beyond Ankur Moitra , Institute for Advanced Studies Wednesday, Februar y 27, 2013, 4:30pm Computer Science 105 Machine learning is a vibrant field with many rich techniques. However, these approaches are often heuristic: we cannot prove good bounds on either their performance or their running time except in limited settings. This talk will focus on designing algorithms whose performance can be analyzed rigorously. As an example of this project, I will describe my work on the nonnegative matrix factorization problem, which has important applications throughout machine learning (and theory). As is often the case, this problem is NP-hard when considered in full generality. However, we introduce a sub-case called separable nonnegative matrix factorization that we believe is the right notion in various contexts. We give a polynomial time algorithm for this problem, and leverage this algorithm to efficiently learn the topics in a Latent Dirichlet Allocation model (and beyond). This is an auspicious example where theory can lead to inherently new algorithms that have highly-practical performance on real data sets. I will also briefly describe some of my other work on learning, including mixtures of Gaussians and robust linear regression, as well as promising directions for future work. Ankur Moitra is an NSF CI Fellow at the Institute for Advanced Study, and also a senior postdoc in the computer science department at Princeton University. He completed his PhD and MS at MIT in 2011 and 2009 respectively, where he was advised by Tom Leighton. He received a George M. Sprowls Award (best thesis) and a William A. Martin Award (best thesis) for his doctoral and master's dissertations. He has worked in numerous areas of algorithms, including approximation algorithms, metric embeddings, combinatorics and smoothed analysis, but lately has been working at the intersection of algorithms and learning.

1 2

Colloquium Speaker John Paisley Mon Feb 25, 4:30pm
by Nicole E. Wagenblast 25 Feb '13

25 Feb '13

Bayesian Nonparametric Models and "Big Data" John Paisley , University of California, Berkeley Monday, February 25, 2013, 4:30pm Computer Science Small Auditorium, Room 105 Bayesian nonparametrics is an area in machine learning in which models grow in size and complexity as data accrue. As such, they they are particularly relevant to the world of "Big Data", where it may be difficult or even counterproductive to fix the number of parameters a priori. A stumbling block for Bayesian nonparametrics has been that their algorithms for posterior inference generally show poor scalability. In this talk, we tackle this issue in the domain of large-scale text collections. Our model is a novel tree-structured model in which documents are represented by collections of paths in an infinite-dimensional tree. We develop a general and efficient variational inference strategy for learning such models based on stochastic optimization, and show that with this combination of modeling and inference approach, we are able to learn high-quality models using millions of documents. John Paisley received the B.S.E. (2004), M.S. (2007) and Ph.D. (2010) in Electrical & Computer Engineering from Duke University, where his advisor was Lawrence Carin. He was a postdoctoral researcher with David Blei in the Computer Science Department at Princeton University, and currently with Michael Jordan in the Department of EECS at UC Berkeley. He works on developing Bayesian models for machine learning applications, particularly for dictionary learning and topic modeling.

1 2

C Park preFPO
by Melissa M. Lawson 21 Feb '13

21 Feb '13

Chris Park will present his preFPO on Tuesday February 26 at 1:30PM in Room 402. The members of his committee are: Olga Troyanskaya, advisor; Mona Singh and Kai Li, readers; Rob Schapire and Moses Charikar, nonreaders. Everyone is invited to attend his talk. His abstract follows below. ---------------------------------------------------------------- Title: Inferring gene function and heterogeneous gene networks from genomic data Abstract With the advancement of quantitative biology, there has been a dramatic increase in functional genomic data. Using these genomic datasets which includes protein-protein interaction, gene expression, localization, and gene knockout studies has enabled researchers to construct functional gene networks. In functional gene networks, each edge represents the probabilistic support for a functional relationship between two genes. However in practice, the goal of biology is to understand further depth of the function and relationship between genes, beyond the abstract notion of a functional relation. In this work, we expand the depth and breadth of functional gene networks. First, we demonstrate that functional gene networks can improve the transfer of gene annotations between organisms to allow prediction algorithms to identify additional genes participating in biological processes that are not already well studied. We apply our novel functional knowledge transfer method to accurately predict novel gene functions across six diverse organisms. Second, biomolecular pathways are built from diverse types of pairwise interactions, ranging from physical protein-protein interactions and modifications to indirect regulatory relationships. One goal of systems biology is to bridge three aspects of this complexity: the growing body of high-throughput data assaying these interactions; the specific interactions in which individual genes participate; and the genome-wide patterns of interactions in a system of interest. Here, we describe methodology for predicting specific types of biomolecular interactions, beyond functional relationship, using high-throughput genomic data in both yeast and human. Ultimately, our work provides further insight into the genetic wiring of the cell enabled by large-scale machine learning with genome-wide biological data.

1 0