Skip to content. Skip to navigation
McGill Home SOCS Home
Personal tools
You are here: Home Announcements and Events Colloquium Archive

Colloquium Home
Fall 2015 Schedule
Winter 2016 Schedule
How to Suggest a Speaker


( Fall 2006 )
Speaker and Abstract
2006/09/08 Speaker: SOCS Professors
Affiliation: McGill University
Area: Computer Science
Title: SOCS Research - Part I
Abstract: This Colloquium, as well as the next one on Sept. 15th, aims at giving a overview of the research being done within the various groups at SOCS. The attendance to both events is mandatory for new M.Sc. students, but all students and profs are invited to hear about what's happening in our department. Biography of Speaker:

2006/09/15 Speaker: SOCS Professors
Affiliation: McGill University
Area: Computer Science
Title: SOCS Research - Part II
Abstract: During this colloquium, every SOCS professor presents an overview of his research in 10 minutes.
2006/09/22 Speaker: Tetsuo Asano
Affiliation: Japan Advanced Institute of Science and Technology
Area: Computational Geometry
Title: Aspect-Ratio Voronoi Diagram with Applications
Abstract: Triangulation is quite important in many areas such as Finete Element Method and computer graphics. In a simple setting we would like to triangulate a simple polygon possibly using a number of internal points. Since a flat or skinny triangle may cause numerical errors, we would like to improve a triangulation if such triangles are included. More precisely, triangles are evaluated by their aspect ratios, which are defined by the ratio between the longest and shortest sides or between the longest side and its corresponding height. Our ultimate goal is to optimize the worst aspect ratio of triangles contained by moving internal points (since polygon vertices are fixed). The problem of achieving the best triangulation in the sense looks quite hard. So, our secondary goal is local improvement of a triangulation. That is, given a triangulation, we find an internal point incident to a triangle of the worst aspect ratio and move it to a locally optimal point in its neighborhood. More precisely, once we find such an internal point p, remove all triangles incident to p. Then, we have a star-shaped polygon (or a convex polygon) around the point. We want to find a point q such that the worst aspect ratio among the triangles resulting by drawing chords from q to all the vertices of the star-shaped polygon is optimized. We can show that such a point is a vertex of a Voronoi diagram, called an aspect-ratio Voronoi diagram which is defined as follows. Given a set of line segments in the plane, a point belongs to a Voronoi region of a line segment if the aspect ratio of a triangle defined by the point and the line segment is largest among all line segments. These Voronoi diagrams are different from an ordinary Voronoi diagram for a point set. After introducing interesting properties, we present three efficient algorithms for finding a point to minimize the largest aspect ratio.
2006/09/29 Speaker: Daniele Micciancio
Affiliation: University of California, San Diego
Title: Cyclic Lattices: Cryptographic Applications and Open Problems
Abstract: Cyclic codes are among the most useful and widely used error correcting codes in coding theory and communication applications. We consider a similarly defined class of "cyclic lattices" (and generalizations), and discuss cryptographic applications, connections with other problems in algebraic number theory, and open problems concerning their computational complexity. Biography of Speaker:

Daniele Micciancio is Associate Professor in the Department of Computer Science and Engineering at the University of California, San Diego. Dr. Micciancio got his MS (1996) and PhD (1998) degrees from MIT, and joined UCSD as a professor in 1999, after spending one more year at MIT as a postdoc. Dr. Micciancio is the recepient of a Matchey Award (FOCS 1998), NSF CAREER award (2001), Hellman fellowship (2001) and Sloan research fellowship (2003). His research interests include computational problems on lattices, coding theory and cryptography. He is the author of the book "Complexity of Lattice problems: a cryptographic perspective".

2006/10/06 Speaker: Janos Pach
Affiliation: City College, CUNY Renyi Institute and NYU
Title: Hypergraph Coloring and Frequency Allocation
Abstract: The following two strongly related coloring problems for geometrically defined hypergraphs arose in connection with frequency allocation in cellular networks. 1. What is the smallest number of colors needed to color the vertices of a hypergraph so that each hyperedge H has a vertex whose color is not shared by any other element of H? 2. Suppose that every hyperedge of a hypergraph is sufficiently large. Is it possible to color the vertices with k colors so that every hyperedge contains at least one point of each color? We give some partial answers to these questions in various settings and ask many open questions. Biography of Speaker:

Janos Pach received his Ph. D. in Mathematics in 1981, from Eotvos University, Budapest, with the Gold Ring of the President of the Hungarian Republic. He is Distinguished Professor of Computer Science at City College, New York, Senior Research Fellow at the Renyi Institute of the Hungarian Academy of Sciences, and Research Professor at Courant Institute, New York University. His main fields of interest are discrete and computational geometry, convexity, and combinatorics. He has written more than 200 research papers. His most recent book ``Research Problems in Discrete Geometry'' with Peter Brass and Willy Moser (McGill) was published last year by Springer. Janos Pach is co-editor-in-chief of Discrete & Computational Geometry and serves on the editorial boards of seven other professional journals. For his scientific achievements, he received the Grunwald Medal from the Janos Bolyai Mathematical Society (1982), the Lester R. Ford Award from the Mathematical Association of America (1990), the Renyi Prize and the Academy Award from the Hungarian Academy of Sciences (1993, 1998). One of his daughters is a freshman at McGill.

2006/10/20 Speaker: Gail Murphy
Affiliation: University of British Columbia
Area: Software Engineering
Title: Reducing Information Overload for Software Developers
Abstract: Software developers are innundated with information: many systems comprise hundreds of thousands of lines or more of source code, their email inboxes are clogged with notifications of new bugs reported and so on. Many of the tools developers use have been engineered to present and deliver this information as fast as possible. The work in our research group aims to reverse this trend by presenting just the information a developer needs when they need it by exploiting structure and patterns in the ways developers work. In this talk, I will present the results of two research projects that take different approaches to this problem. In the Mylar project, we are using patterns in task-based interaction to focus the user interface of an integrated development environment. In the Sybil project, we are using a recommendation approach to ease the task of bug triage. Biography of Speaker:

Gail Murphy is a Professor in the Department of Computer Science at the University of British Columbia. She joined UBC in 1996 after completing Ph.D. and M.S. degrees at the University of Washington. Before returning to graduate school, she worked as a software engineer at a research and development telecommunications company for 5 years. She also holds a B.Sc. degree from the University of Alberta. She works primarily on building simpler and more effective tools to help developers manage software evolution tasks. In 2005, she held a UBC Killam Research Fellowship and also received the Dahl-Nygaard Junior Prize for her work in software evolution from AITO. In 2006 she received an NSERC Steacie Fellowship and the CRA-W Anita Borg Early Career Award. She has recently been a co-author on two papers awarded SIGSOFT distinguished paper awards. One of the most rewarding parts of her career has been collaborating with many talented graduate and undergraduate students.

2006/11/03 Speaker: Stephen Hsu
Affiliation: University of Oregon
Title: Introduction to Technology Startups
Abstract: 20 years ago applied research and development was concentrated at corporate labs like Bell, IBM, Xerox PARC, etc. Today, innovation is more likely to be found at small, venture capital backed companies founded by creative risk takers. The odds have never been greater that you, a scientist or engineer, might someday work at (or found!) a startup company. This talk is an introduction to this important and dynamic part of our economy, from the perspective of a professor and serial entrepreneur. Biography of Speaker:

Dr. Stephen D. H. Hsu is a professor of theoretical physics at the University of Oregon. He received a Ph.D. from the University of California, Berkeley in 1991 and B.S. from the California Institute of Technology in 1986. He was a Harvard Junior Fellow and assistant professor at Yale University before moving to Oregon in 1998. He is a co-founder of SafeWeb, which was acquired by Symantec on October 15, 2003. In 2005 Stephen, along with SafeWeb co-founder James Hormuzdiar, co-founded Robotgenius, an Oakland based security software company.

2006/11/17 Speaker: Daniel Levitin
Affiliation: McGill University
Title: Understanding Musical Preferences: Perspectives from Cognitive Science
Abstract: In 2006 so far, more than 1,000 times as many music files were downloaded illegally as legally. This has sent analysts and industry executives scurrying to find ways to remonetize the music industry (and it has sent film company executives running for the smelling salts). Over 20 start-ups around the world have been tackling the problem of developing robust music recommendation engines as a way to charge consumers for music (by charging for the recommendation service), but so far, none have been as successful as a good disc jockey at creating coherent, interesting playlists. I will review the work I did as the chief architect of, the first internet music recommendation system (which just sold to the AMG Group), followed by my work on the second generation music recommendation system, MusicGenome (now incorporated in, up through the fourth generation recommendation system that McGill licensed to Astral Media for last spring. One problem with music recommendation is that musical preferences are multi-dimensional, describing a 60-D non-Euclidean manifold. Sound examples will be included. Biography of Speaker:

Daniel Levitin, Ph.D. is the Bell Chair in the Psychology of Electronic Communication at McGill, with a principal appointment in the Department of Psychology. He holds an Associate membership in the SOCS, and in the Schulich School of Music. Prior to entering academia, he was an award-winning record producer, and served as Vice-President of talent acquisition (and later President) of 415/Columbia Records, which was sold to Sony Music in 1990. He worked for Microsoft co-founder Paul Allen's Interval Research (a Silicon Valley Think-Tank) during the 1990s, developing the ideas that eventually became the first music downloading and recommendation dot-com, MoodLogic, and working with computer industry pioneers including David Liddle (Xerox Alto and Star), Joy Mountford (Apple Computer Desktop design), Malcolm Slaney (Apple Computer sound tools), and Dick Shoup (the laser printer). He has authored 50 scientific articles on music cognition, and statistical methods for analyzing acoustical data (including spherical statistics and functional data analysis). Prior to joining McGill, he taught Computer Science at Stanford University. He is the author of the best-selling book "This Is Your Brain on Music: The Science of a Human Obsession."