All Seminars
| Title: Evaluating E-Discovery Search |
|---|
| Colloquium: Computer Science |
| Speaker: Douglas W. Oard of University of Maryland |
| Contact: Eugene Agichtein, eugene@mathcs.emory.edu |
| Date: 2011-09-20 at 10:00AM |
| Venue: White Hall 208 |
| Download Flyer |
| Abstract: Civil litigation in this country relies on each side making relevant evidence available to the other, a process known as "discovery." The explosive growth of information in digital form has led to an increasing focus on how search technology can best be applied to balance costs and responsiveness in what has come to be known as "e-discovery". This is now a multi-billion dollar business, one in which new vendors are entering the market frequently, usually with impressive claims about the efficacy of their products or services. Courts, attorneys, and companies are actively looking to understand what should constitute best practice, both in the design of search technology and in how that technology is employed. In this talk I will provide an overview of the e-discovery process, and then I will use that background to motivate a discussion of which aspects of that process the TREC Legal Track is seeking to model. I will then spend most of the talk describing two novel aspects of evaluation design: (1) recall-focused evaluation in large collections, and (2) modeling an interactive process for "responsive review" with fairly high fidelity. Although I will draw on the results of participating teams to illustrate what we have learned, my principal focus will be on discussing what we presently understand to be the strengths and weaknesses of our evaluation designs.\\ \\ Bio:\\ Douglas Oard is a Professor at the University of Maryland, College Park, with joint appointments in the College of Information Studies and the Institute for Advanced Computer Studies, where he is the director of the Computational Linguistics and Information Processing Lab. Dr. Oard earned his Ph.D. in Electrical Engineering from the University of Maryland, and his research interests center around the use of emerging technologies to support information seeking by end users. His recent work has focused on interactive techniques for cross-language information retrieval, searching conversational media such as speech and email, evaluation design for e-discovery in the TREC Legal Track, and support for sense-making in large digital archival collections. Additional information is available at http://terpconnect.umd.edu/~oard/. |
| Title: Nonnegative Polynomials and Sums of Squares |
|---|
| Seminar: Algebra and Number Theory |
| Speaker: Greg Blehkerman of Georgia Tech |
| Contact: Vicki Powers, vicki@mathcs.emory.edu |
| Date: 2011-09-20 at 3:00PM |
| Venue: MSC E406 |
| Download Flyer |
| Abstract: A multivariate real polynomial $p$ is nonnegative if $p(x)$ is at least $0$ for all $x \in \mathbb{R}^n$. I will review the history and motivation behind the problem of representing nonnegative polynomials as sums of squares. Such representations are of interest for both theoretical and practical computational reasons. However, much about the relationship between nonnegative polynomials and sums of squares remains unknown. I will present some recent results which establish links between nonnegative polynomials, sums of squares and classical algebraic geometry. In the smallest cases where there exist nonnegative polynomials that are not sums of squares I will describe a complete classification of the differences between these sets. |
| Title: A solution to a problem of determining the sides of a lens |
|---|
| Seminar: Analysis and Differential Geometry |
| Speaker: Hasan Palta of Emory University |
| Contact: Vladimir Oliker, oliker@mathcs.emory.edu |
| Date: 2011-09-20 at 4:00PM |
| Venue: MSC W301 |
| Download Flyer |
| Abstract: Suppose that a beam of light with the positive vertical direction $\bf{k}$ is crossing a domain $\Omega$ in the horizontal plane $z=0$ with some intensity $I\in L^1(\bar{\Omega})$ and is refracted at both sides of a lens in such a way that the final direction is also $\bf{k}$ and that the beam illuminates a set $T_d$ in the plane $z=d$ with intensity $L\in L^1(\bar{T}_d)$. Let $n_1$ and $n_2$ be the refractive indices of the ambient environment and of the lens, respectively. Such a construction generates a mapping $P:\Omega\to T$ where $T$ is the orthogonal projection of the domain $T_d$ onto $z=0$. We consider the inverse problem of recovering the two sides $z\in C(\bar{\Omega})$ and $w\in C(\bar{T})$ of the lens for given domains $\Omega$ and $T_d$ and the corresponding intensities $I$ and $L$. In analytic formulation, this problem requires a solution to a nonlinear partial differential equation of Monge-Amp\`{e}re type. In this talk, we present a different approach to this problem, describe an algorithm giving approximate solutions using general properties of geometric optics and give some examples. |
| Title: Improving Question-Answering Using Human-Machine Synergies |
|---|
| Seminar: Computer Science |
| Speaker: Ryen White of Microsoft |
| Contact: Eugene Agichtein, eugene@mathcs.emory.edu |
| Date: 2011-09-09 at 3:00PM |
| Venue: MSC W301 |
| Download Flyer |
| Abstract: Question-and-answer (Q and A) has received significant recent attention with the success of IBM Watson on the quiz show Jeopardy!. Machine-based Q and A is promising, but the technology is some way from rivaling the abilities of humans, even in seemingly rudimentary tasks such as question comprehension and clarification dialog. In this talk, I will argue that by leveraging human-machine synergies, we can create powerful mediated Q and A systems that can overcome the limitations of machine-only solutions, while also meeting basic human desires for interaction with others. Specifically, I will: (1) describe our work on developing IM-an-Expert, a synergistic Q and A system recently shipped by Microsoft Lync; (2) present findings from two studies: one investigating the effect of community size and contact rate on the effectiveness of real-time social Q and A, and another examining the value of predicting outcomes at various points of the question lifecycle, and; (3) provide insights and lessons learned from our deployment of real-time Q and A technology within Microsoft and beyond.\\ \\ Bio:\\ \\ Ryen W. White is a researcher in the Context, Learning, and User Experience for Search (CLUES) Group at Microsoft Research, Redmond. His research interests lie in understanding search interaction and in developing tools to help people search more effectively. He received his Ph.D. in Interactive Information Retrieval from the Department of Computing Science, University of Glasgow, United Kingdom, in 2004. Ryen has published over 100 conference papers and journal articles in Web search, log analysis, and user studies of search systems. He has received five best-paper awards, including two at the ACM SIGIR conference (2007,2010), one at the ACM SIGCHI conference (2011), and one in JASIST (2010). His doctoral research received the British Computer Society’s Distinguished Dissertation Award for the best Computer Science Ph.D. dissertation in the United Kingdom in 2004/2005. Ryen has co-organized numerous workshops on information seeking, in particular exploratory search, including an NSF-sponsored invitational workshop, and has guest co-edited special issues in these areas for a variety of outlets, including Communications of the ACM and IEEE Computer. Since 2008, he has co-organized the annual HCIR workshop. Ryen has served as area chair for many top conferences in information retrieval, and currently serves on the editorial board of ACM TOIS and the Information Retrieval Journal. In addition to academic impact, his research has been shipped in many Microsoft products, including Bing, Xbox, Internet Explorer, and Lync. |
| Title: Mock period functions, sesquiharmonic Maass forms, and non-critical values of L-functions |
|---|
| Seminar: Algebra |
| Speaker: Kathrin Bringmann of University of Cologne |
| Contact: Zachary A. Kent, kent@mathcs.emory.edu |
| Date: 2011-09-06 at 3:00PM |
| Venue: MSC E406 |
| Download Flyer |
| Abstract: In this talk we view the generating functions of L-functions associated to cusp forms in the framework of automorphic forms. We prove that it can be interpreted as a "mock period function". Further, we show that non-critical values can be encoded into a "sesquiharmonic Maass form". Finally, we formulate and prove an Eichler-Shimura-type isomorphism for the space of mock period functions. |
| Title: Symmetric chain decompositions of quotients of partially ordered sets |
|---|
| Seminar: Combinatorics |
| Speaker: Dwight Duffus of Emory University |
| Contact: Dwight Duffus, dwight@mathcs.emory.edu |
| Date: 2011-09-02 at 4:00PM |
| Venue: W306 |
| Download Flyer |
| Abstract: Given a subgroup G of the automorphism group of a partially ordered set P, the quotient P/G has as its elements the orbits in P under G with ordering induced by that of P. Canfield and Mason have suggested that for the Boolean lattice of all subsets of a finite set and any subgroup of its automorphism group, these quotients are symmetric chain orders. With Jeremy McKibben-Sanders and Kyle Thayer, we have shown this to be true in some special cases. |
| Title: Pascal's rule for vector bundles of rank four |
|---|
| Seminar: Algebra and number theory |
| Speaker: Asher Auel of Emory University |
| Contact: R. Parimala, parimala@mathcs.emory.edu |
| Date: 2011-08-30 at 3:00PM |
| Venue: MSC E406 |
| Download Flyer |
| Abstract: The low dimensional ``exceptional'' isomorphisms of Dynkin diagrams have been exploited in various ways to describe torsors for the corresponding linear algebraic groups over fields and rings. In this talk, we show how a version of the combinatorial Pascal's rule, adapted to vector bundles over algebraic varieties, can be used to define a Witt group-theoretic obstruction to the existence of certain decompositions. For vector bundles of rank four over projective varieties, we use the exceptional isomorphism $A_3=D_3$ to prove that this is the only obstruction. |
| Title: The partition function modulo prime powers |
|---|
| Seminar: Algebra and Number Theory |
| Speaker: Matthew Boylan of University of South Carolina |
| Contact: Zachary A. Kent, kent@mathcs.emory.edu |
| Date: 2011-08-25 at 3:00PM |
| Venue: MSC E406 |
| Download Flyer |
| Abstract: Recently, Folsom, Kent, and Ono discovered surprising general arithmetic properties of values of p(n), the ordinary partition function, modulo prime powers. More precisely, let l \textgreater 3 be prime, and let m be a positive integer. Their work implies systematic linear relations modulo l\^{}m among values of p(n) in certain arithmetic progressions modulo l\^{}b for all odd b \textgreater b (l,m), a constant depending on l and m. In this talk, we prove a refined upper bound on b(l,m). Our bound is sharp in all computed cases. Abstractly, b(l,m) measures the stabilization rate of a certain sequence of modules of modular forms with coefficients reduced modulo l\^{}m. To define these modules, Folsom, Kent, and Ono introduce a new operator, D(l). We obtain our bound by carefully studying how D(l) effects filtrations of the relevant modular forms. This is joint work with John Webb. |
| Title: On the Number of Edges in 2-factor Isomorphic Graphs |
|---|
| Defense: Dissertation |
| Speaker: Paul Wrayno of Emory University |
| Contact: Paul Wrayno, pwaryno@emory.edu |
| Date: 2011-06-30 at 4:00PM |
| Venue: W306 |
| Download Flyer |
| Abstract: A 2-factor is a collection of disjoint cycles in a graph that cover all vertices of that graph. A graph is called 2-factor isomorphic if all of its 2-factors are the same when viewed as a multiset of unlabeled cycles.\\ \\ In this dissertation, we find the maximum size of 2-factor isomorphic graphs that contain a desired 2-factor. We are also able to give general bounds when no 2-factor is specified or any 2-factor with a fixed number of cycles is desired. We also find similar results for the special case where the underlying graph is bipartite. In each case we provide constructions that attain the maximum size. |
| Title: Large-Scale Inverse Problems in Imaging: Two Case Studies |
|---|
| Defense: Computer Science |
| Speaker: Sarah Knepper of Emory University |
| Contact: Sarah Knepper, smknepp@emory.edu |
| Date: 2011-06-06 at 11:00AM |
| Venue: W302 |
| Download Flyer |
| Abstract: Solving inverse problems is an important part of scientific computing. As computers become more powerful, solutions to increasingly larger problems are sought, allowing for more accurate representations of real-world applications. We consider solving large-scale inverse problems, ranging from linear to fully nonlinear. We look at aspects common to inverse problems, such as their ill-posedness, and see how regularization can help produce meaningful results. We discuss a number of different methods for solving while providing regularization. One such technique is to solve using an iterative method but stop the iterations early, before convergence is fully achieved. Iterative solvers are particularly useful for large-scale inverse problems as computations can be done in parallel. Trilinos is a mathematical software library for solving problems coming from many fields of scientific computing. One particular package, Belos, provides both an abstract framework and concrete implementations of various iterative solvers. We have implemented two additional solvers within the Belos framework, LSQR and MRNSD, which can be used to solve linear inverse problems.\\ \\ We then consider two different case studies, where we wish to solve a large-scale linear inverse problem. In the first study, we want to remove patient motion blur from positron emission tomography (PET) images when motion information is tracked and recorded during the scan. We describe how this problem can be formulated as a linear equation, then we solve it using the solvers we implemented. We also look at a number of results, seeing how the reconstruction improves as more motion information is included in our model. The second case study comes from the field of adaptive optics. Here we wish to determine the distortion caused by the atmosphere when imaging using ground-based telescopes. Sensors are able to obtain noisy estimates of the gradients of the distortion, resulting in a Kronecker product-structured linear least squares problem. We describe a solving method that employs Tikhonov-type regularization by exploiting properties of the Kronecker product and utilizing the generalized singular value decomposition (GSVD). Our approach includes constructing a preconditioner off-line and then applying a few iterations of preconditioned LSQR. |