From cnovak at tti-c.org Mon Mar 3 11:45:50 2008 From: cnovak at tti-c.org (Christina Novak) Date: Mon Mar 3 07:03:56 2008 Subject: [TTIC Colloquium] TTI-C Distinguished Lecture Series MAR 6- Avi Widgerson (IAS Princeton University) Message-ID: <00a001c87d56$6dc02eb0$a9bf8780@cnovakHBRQFD1> Good afternoon, The Toyota Technological Institute at Chicago would like to invite you to our 2008 Distinguished Lecture Series on the University of Chicago Campus. Our first speaker will be Dr. Avi Widgerson. (Details, see http://tti-c.org/dls) The lecture will be held at 3:30pm Biological Sciences Learning Center (room 115- 1st floor) 924 East 57th St. Chicago, IL 60637 Thursday, March 6th, 2008 "Randomness: a computational complexity view" Avi Widgerson (Institute for Advanced Study, Princeton University) ABSTRACT: Man has grappled with the meaning and utility of randomness for centuries. Research in the Theory of Computation in the last thirty years has enriched this study considerably. I'll describe two main aspects of this research on randomness, demonstrating its power and weakness respectively. -Randomness is paramount to computational efficiency: The use of randomness can dramatically enhance computation (and do other wonders) for a variety of problems and settings. In particular, examples will be given of probabilistic algorithms (with tiny error) for natural tasks in different areas of mathematics, which are exponentially faster than their (best known) deterministic counterparts. -Computational efficiency is paramount to understanding randomness: I will explain the computationally-motivated definition of "pseudorandom" distributions, namely ones which cannot be distinguished from the uniform distribution by efficient procedure from a given class. We then show how such pseudorandomness may be generated deterministically, from (appropriate) computationally difficult problems. Consequently, randomness is probably not as powerful as it seems above. I'll conclude with the power of randomness in other computational settings, primarily probabilistic proof systems. We discuss the remarkable properties of Zero-Knowledge proofs and of Probabilistically Checkable proofs. We look forward to seeing you at the lecture! Sincerely, David McAllester TTI-C Chief Academic Officer -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20080303/4368eebd/attachment-0001.htm From macglashan at tti-c.org Tue Mar 4 08:33:49 2008 From: macglashan at tti-c.org (Julia MacGlashan) Date: Tue Mar 4 03:43:08 2008 Subject: [TTIC Colloquium] TTI-C Talk Tomorrow: Parikshit Gopalan, University of Washington Message-ID: <000301c87e04$c39f5ff0$aabf8780@jmacglDPLFYD1> When: Wed, Mar 5, 2008 @ 10:00 am Where: TTI-C Conference Room Who: Parikshit Gopalan, University of Washington Topic: Fitting Polynomials to Noisy Data The problem of finding the polynomial that best fits a noisy data-set (or polynomial reconstruction) has a long history, dating back to curve-fitting problems studied in the 1800s. In the last two decades, there has been tremendous progress on this problem in computer science, driven by the discovery of powerful new algorithms. These results have spurred exciting new developments in Coding theory, Computational learning, Cryptography and Hardness of Approximation. In this talk, we will explore this problem from the perspectives of Coding theory and Computational learning. We begin with an algorithm for decoding a well-studied family of binary error-correcting codes called Reed-Muller codes, which are obtained from low-degree polynomials. The salient feature of this algorithm is that it works even when the number of errors far exceeds the so-called Johnson bound. I will present an algorithm for agnostically learning decision trees under the uniform distribution. This is the first polynomial time algorithm for learning decision trees in a harsh noise model. This algorithm solves the reconstruction problem for real polynomials using tools from convex optimization. I will also discuss settings where the reconstruction problem seems intractable. We will see evidence that the notorious Noisy Parity problem is hard under the uniform distribution. We will present hardness results suggesting that learning simple concepts with noise is impossible for arbitrary distributions. Contact: Julia Chuzhoy, TTI-C cjulia@tti-c.org 4-2490 About Parikshit Gopalan: Parikshit Gopalan grew up in India in the city of Bombay (now called Mumbai). He graduated with an undergraduate degree from IIT-Bombay (whose name, thankfully, has not changed). He received his Ph.D from Georgia Tech in August 2006, under the guidance of Dick Lipton. Following this, he did a short stint as a postdoctoral researcher at the University of Texas at Austin. He is currently a postdoc at the University of Washington, visiting Princeton University. His research focuses on theoretical computer science, especially on algebraic problems arising from algorithms and complexity. He also likes to dabble in other areas such as Data-stream algorithms and Communication complexity. His website is: http://www.cs.washington.edu/homes/parik -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20080304/e02c3714/attachment.htm From macglashan at tti-c.org Wed Mar 5 09:23:56 2008 From: macglashan at tti-c.org (Julia MacGlashan) Date: Wed Mar 5 04:33:13 2008 Subject: [TTIC Colloquium] TTI-C Talk: Ryan Williams, Carnegie Mellon University Message-ID: <001701c87ed4$edf4a960$aabf8780@jmacglDPLFYD1> When: Tue, Mar 11, 2008 @ 2:00 pm Where: TTI-C Conference Room Who: Ryan Williams, Carnegie Mellon University Topic: Applying Practice to Theory: Time Lower Bounds for Fundamental Problems A fertile area of recent research has demonstrated concrete polynomial time lower bounds for solving natural hard problems on restricted computational models. Among these problems are Satisfiability, Vertex Cover, Hamilton Path, MOD6-SAT, Majority-of-Majority-SAT, and Tautologies, to name a few. These lower bound proofs all follow a certain diagonalization-based proof-by-contradiction strategy. A pressing open problem has been to determine how powerful such proofs can possibly be. I will survey some of my work in this area, with an emphasis on simplicity. After a brief overview of the proof techniques used, I will propose an automated search strategy for studying the limits of these proof techniques. In particular, I will demonstrate how the search for better lower bounds can often be turned into a problem of solving a large series of linear programming instances. This mathematical work has in turn led to a new methodology for proving time lower bounds, where prospective lower-bounders formulate their proof rules, write programs to generate optimal short proofs, then study the empirical results and extrapolate new proofs on paper. In some settings, our programs provide strong evidence that the best known lower bound proofs are already optimal for the current framework, contradicting the consensus intuition; in other settings, we are guided to improved lower bounds where no further progress had been made for some time. Contact: Lance Fortnow fortnow@eecs.northwestern.edu 834-9873 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20080305/073bdc8e/attachment.htm From macglashan at tti-c.org Mon Mar 10 08:04:46 2008 From: macglashan at tti-c.org (Julia MacGlashan) Date: Mon Mar 10 08:04:56 2008 Subject: [TTIC Colloquium] TTI-C Talk: Ryan Williams, Carnegie Mellon University Message-ID: <000301c882b7$b4481cb0$aabf8780@jmacglDPLFYD1> When: Tue, Mar 11, 2008 @ 2:00 pm Where: TTI-C Conference Room Who: Ryan Williams, Carnegie Mellon University Topic: Applying Practice to Theory: Time Lower Bounds for Fundamental Problems A fertile area of recent research has demonstrated concrete polynomial time lower bounds for solving natural hard problems on restricted computational models. Among these problems are Satisfiability, Vertex Cover, Hamilton Path, MOD6-SAT, Majority-of-Majority-SAT, and Tautologies, to name a few. These lower bound proofs all follow a certain diagonalization-based proof-by-contradiction strategy. A pressing open problem has been to determine how powerful such proofs can possibly be. I will survey some of my work in this area, with an emphasis on simplicity. After a brief overview of the proof techniques used, I will propose an automated search strategy for studying the limits of these proof techniques. In particular, I will demonstrate how the search for better lower bounds can often be turned into a problem of solving a large series of linear programming instances. This mathematical work has in turn led to a new methodology for proving time lower bounds, where prospective lower-bounders formulate their proof rules, write programs to generate optimal short proofs, then study the empirical results and extrapolate new proofs on paper. In some settings, our programs provide strong evidence that the best known lower bound proofs are already optimal for the current framework, contradicting the consensus intuition; in other settings, we are guided to improved lower bounds where no further progress had been made for some time. Contact: Lance Fortnow fortnow@eecs.northwestern.edu 834-9873 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20080310/29770440/attachment-0001.htm From cnovak at tti-c.org Mon Mar 10 14:43:10 2008 From: cnovak at tti-c.org (Christina Novak) Date: Mon Mar 10 14:43:21 2008 Subject: [TTIC Colloquium] TTI-C Talk- Dorit Hochbaum (UC Berkeley) Fri. 3/14 Message-ID: <010f01c882ef$5acc1d20$a9bf8780@cnovakHBRQFD1> Dorit S. Hochbaum, Haas School of Business and Department of IE&OR University of California, Berkeley Fri Mar 14 2:30pm TTI-C Conference Room "Efficient cut-based image segmentation techniques" Segmenting an image is to determine a partition to the salient features of the image and identify them as associated with different types of objects. This is of particular importance in medical imaging where blur conceals information of critical importance. The MRF presentation of the problem is formulated as minimization of deviation penalty, from the captured colors of the pixels, and separation penalty, which is associated with two adjacent pixels having different colors. We describe a very efficient and best possible polynomial time algorithm for a convex variant of the problem. This algorithm's efficiency enables its use in an interactive set-up. It is more efficient than most procedures based on spectral techniques, partitioning approaches or heuristic clustering. We then demonstrate how to apply the procedure for the purpose of de-blurring medical images and identifying structures hidden by noise. Time permitting, we will present additional efficient poly time algorithms for several types of ratio cuts that have been believed to be "hard". Any questions, please contact Chrissy Novak cnovak@tti-c.org or (773)834-2216 or Ronen Basri (773) 834-2515. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20080310/38303754/attachment.htm From macglashan at tti-c.org Thu Mar 13 13:58:05 2008 From: macglashan at tti-c.org (Julia MacGlashan) Date: Thu Mar 13 13:58:16 2008 Subject: [TTIC Colloquium] TTI-C Talk: Ira Kemelmacher-Shlizerman, Weizmann Institute of Science Message-ID: <003a01c88544$8e8832a0$aabf8780@jmacglDPLFYD1> When: Tuesday, March 25, 2008 @ 2:30 pm Where: TTI-C Conference Room (2nd Floor Lobby, Rm. #230) Who: Ira Kemelmacher-Shlizerman, Weizmann Institute of Science Topic: 3D Shape Reconstruction from gray-level and two-tone images Lighting has a significant effect on the appearance of objects. At the same time lighting provides information that can be used to recover the 3-dimensional shape of objects. It is long known that recovering shape from shading information is ill-posed, yet people appear to have a good sense of the underlying three-dimensional world already from a single picture. In this talk I will show how explicit modeling of lighting, along with the incorporation of prior knowledge, can facilitate 3D reconstruction. In the first part of my talk I will introduce a novel algorithm for 3D shape recovery of faces from single images using a single 3D reference model of a different person's face. The method uses the input image as a guide to mold the reference model to reach a desired reconstruction. Assuming Lambertian reflectance and rough alignment of the input image and reference model, we seek shape, albedo, and lighting that best fit the image while preserving the rough structure of the model. We demonstrate our method by providing accurate reconstructions of novel faces overcoming significant differences in shape due to gender, race, and facial expressions. I will then discuss 3D reconstruction from two-tone images (Mooney images). These images are fascinating testimonial to the ability of biological vision systems to accurately handle and interpret impoverished data. Their ambiguous nature, face specificity, and sudden interpretability have fascinated psychologists and neurobiologists throughout the past half a century and led to a flurry of studies. Our analysis indicates that reconstruction from such images is very ambiguous even if we consider only reconstruction along the "Mooney transition curve," the boundary curve between black and white pixels. However with prior knowledge it is possible to successfully recover the 3D shape of faces from single Mooney images. This is joint work with Ronen Basri. Parts of the work were done in collaboration with Boaz Nadler (Weizmann) and David W. Jacobs (Maryland). Contact: Ronen Basri, TTI-C ronen.basri@tti-c.org 834-2515 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20080313/3cd0b65d/attachment.htm From macglashan at tti-c.org Thu Mar 20 09:42:33 2008 From: macglashan at tti-c.org (Julia MacGlashan) Date: Thu Mar 20 09:42:15 2008 Subject: [TTIC Colloquium] Date Change: Ira Kemelmacher-Shlizerman, Weizmann Institute of Science @ TTI-C Message-ID: <000b01c88aa1$05edd6e0$aabf8780@jmacglDPLFYD1> When: Wednesday, March 26, 2008 @ 10:00 am Where: TTI-C Conference Room Who: Ira Kemelmacher-Shlizerman, Weizmann Institute of Science Topic: 3D Shape Reconstruction from gray-level and two-tone images Lighting has a significant effect on the appearance of objects. At the same time lighting provides information that can be used to recover the 3-dimensional shape of objects. It is long known that recovering shape from shading information is ill-posed, yet people appear to have a good sense of the underlying three-dimensional world already from a single picture. In this talk I will show how explicit modeling of lighting, along with the incorporation of prior knowledge, can facilitate 3D reconstruction. In the first part of my talk I will introduce a novel algorithm for 3D shape recovery of faces from single images using a single 3D reference model of a different person's face. The method uses the input image as a guide to mold the reference model to reach a desired reconstruction. Assuming Lambertian reflectance and rough alignment of the input image and reference model, we seek shape, albedo, and lighting that best fit the image while preserving the rough structure of the model. We demonstrate our method by providing accurate reconstructions of novel faces overcoming significant differences in shape due to gender, race, and facial expressions. I will then discuss 3D reconstruction from two-tone images (Mooney images). These images are fascinating testimonial to the ability of biological vision systems to accurately handle and interpret impoverished data. Their ambiguous nature, face specificity, and sudden interpretability have fascinated psychologists and neurobiologists throughout the past half a century and led to a flurry of studies. Our analysis indicates that reconstruction from such images is very ambiguous even if we consider only reconstruction along the "Mooney transition curve," the boundary curve between black and white pixels. However with prior knowledge it is possible to successfully recover the 3D shape of faces from single Mooney images. This is joint work with Ronen Basri. Parts of the work were done in collaboration with Boaz Nadler (Weizmann) and David W. Jacobs (Maryland). Contact: Ronen Basri, TTI-C ronen.basri@tti-c.org 834-2515 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20080320/317ddc49/attachment-0001.htm From macglashan at tti-c.org Mon Mar 24 14:57:45 2008 From: macglashan at tti-c.org (Julia MacGlashan) Date: Mon Mar 24 14:57:06 2008 Subject: [TTIC Colloquium] Ira Kemelmacher-Shlizerman, Weizmann Institute of Science: TTI-C Talk Message-ID: <005601c88df1$b609a7e0$aabf8780@jmacglDPLFYD1> When: Wednesday, March 26, 2008 @ 10:00 am Where: TTI-C Conference Room Who: Ira Kemelmacher-Shlizerman, Weizmann Institute of Science Topic: 3D Shape Reconstruction from gray-level and two-tone images Lighting has a significant effect on the appearance of objects. At the same time lighting provides information that can be used to recover the 3-dimensional shape of objects. It is long known that recovering shape from shading information is ill-posed, yet people appear to have a good sense of the underlying three-dimensional world already from a single picture. In this talk I will show how explicit modeling of lighting, along with the incorporation of prior knowledge, can facilitate 3D reconstruction. In the first part of my talk I will introduce a novel algorithm for 3D shape recovery of faces from single images using a single 3D reference model of a different person's face. The method uses the input image as a guide to mold the reference model to reach a desired reconstruction. Assuming Lambertian reflectance and rough alignment of the input image and reference model, we seek shape, albedo, and lighting that best fit the image while preserving the rough structure of the model. We demonstrate our method by providing accurate reconstructions of novel faces overcoming significant differences in shape due to gender, race, and facial expressions. I will then discuss 3D reconstruction from two-tone images (Mooney images). These images are fascinating testimonial to the ability of biological vision systems to accurately handle and interpret impoverished data. Their ambiguous nature, face specificity, and sudden interpretability have fascinated psychologists and neurobiologists throughout the past half a century and led to a flurry of studies. Our analysis indicates that reconstruction from such images is very ambiguous even if we consider only reconstruction along the "Mooney transition curve," the boundary curve between black and white pixels. However with prior knowledge it is possible to successfully recover the 3D shape of faces from single Mooney images. This is joint work with Ronen Basri. Parts of the work were done in collaboration with Boaz Nadler (Weizmann) and David W. Jacobs (Maryland). Contact: Ronen Basri, TTI-C ronen.basri@tti-c.org 834-2515 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20080324/9c22ee6f/attachment.htm From cnovak at tti-c.org Thu Mar 27 08:26:31 2008 From: cnovak at tti-c.org (Christina Novak) Date: Thu Mar 27 08:30:04 2008 Subject: [TTIC Colloquium] TTI-C Distinguished Lecture Series APRIL 1- Prabhakar Raghavan (Yahoo! Research) Message-ID: <003a01c89016$90398c70$a9bf8780@cnovakHBRQFD1> Good morning, The Toyota Technological Institute at Chicago would like to invite you to our 2008 Distinguished Lecture Series on the University of Chicago Campus. Our second speaker in the series will be Dr. Prabhakar Raghavan. (Details, see http://tti-c.org/dls) The lecture will be held at 3:30pm Biological Sciences Learning Center (room 109- 1st floor) 924 East 57th St. Chicago, IL 60637 Tuesday, April 1st, 2008 "New sciences for a new web" Prabhakar Raghavan (Yahoo! Research) ABSTRACT: The web has made a widely-hailed transition from its original incarnation to a putative state of "Web 2.0". This transition has stemmed from the clever use of AJAX and efficient grid computing to enhance a user's perception of responsiveness and interaction. In the process, the web experience has changed from a human interacting with a browser, to the emergence of a plethora of social media experiences. One consequence is that moving beyond the current notion of Web 2.0 demands research advances that straddle the boundaries between computational and social sciences, the latter including microeconomics, cognitive psychology and sociology. It also raises difficult questions on the use of data - ranging from the algorithmic to the societal. This lecture will attempt to chart this interdisciplinary research agenda, arguing that the most influential research will require heavy interaction between these "hard" and "soft" sciences. We look forward to seeing you at the lecture. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20080327/c680929a/attachment.htm