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

Colloquium Home
Fall 2014 Schedule
Winter 2015 Schedule
How to Suggest a Speaker
Archives

 

Date
( Winter 2008 )
Speaker and Abstract
2008/02/01 Speaker: Jason Della Rocca
Affiliation: International Game Developers Association
Title: Graphs to Grok Games
Abstract: From the Uncanny Valley to the Innovator's Dilemma this lecture will make extensive use of graphs and charts (ya, it's more exciting than it sounds, really) as a visual aid to understanding the vibrancy of the game industry ecosystem. Topics will range from current technical trends and challenges, to structural business issues and the everyday game industry professional. Biography of Speaker:

Jason is the executive director of the International Game Developers Association (IGDA), a professional society committed to advancing the careers and enhancing the lives of game developers. Jason and the IGDA focus on connecting developers with their peers, promoting professional development, and advocating on issues that affect the developer community -- such as quality of life, creative freedoms, workforce diversity and credit standards. As the spokesperson for the IGDA, Jason has appeared in countless news outlets (e.g., Wired, Nightline, LA Times, NPR, Wall Street Journal, G4, etc) and has spoken at conferences around the world (e.g., GDC, E3, TGS, SIGGRAPH, ChinaJoy, DiGRA, etc). Jason has been a member of the game development community for over a decade, and has spent time at Matrox Graphics, Quazal and Silicon Graphics. He blogs at www.realitypanic.com.


2008/02/22 Speaker: Anne Condon
Affiliation: University of British Columbia
Area: Bioinformatics
Title: Improving Nucleic Acid Secondary Structure Prediction Via Refined Models
Abstract: RNA molecules are increasingly in the spotlight, in recognition of their important regulatory and catalytic roles in the cell and their promise in therapeutics. Outside of the cell, DNA nanostructures are now finding use in the construction of biosensors, nanotubes, lattices.
Function follows form in the molecular world, and so our ability to understand nucleic acid function and structure, and design novel structures, is enhanced by reliable means for secondary structure prediction. Methods that predict structure from the base sequence find that structure with minimum free energy, with respect to a standard energy model developed by biochemists. However, the quality of predictions is limited by the quality of the energy model, and MFE predictions are not informed by the folding pathway.
In this talk, we'll describe two steps towards more accurate structure prediction. The first is improved inference of thermodynamic parameters, informed by a large repository of known structures and using maximum likelihood techniques. The second is the use of a two-phase approach to pseudoknotted structure prediction, which models hierarchical folding. Biography of Speaker:

Anne Condon is a Professor of Computer Science at U. British Columbia and Associate Dean of Science for Faculty Affairs and Strategic Initiatives.
Anne's research currently focuses on computational prediction of RNA and DNA structure, with applications to design of novel structures and gene synthesis. Her research contributions span computational complexity theory, hardware verification, biomolecular computation, and combinatorial auctions.
Anne received her B.Sc. degree (1982) from University College Cork, Ireland and her Ph.D. (1987) from the University of Washington, Seattle. She was was a faculty member at U. Wicsonsin at Madison from 1987-1999. She won an ACM Distinguished Dissertation award, an NSF National Young Investigator Award, and the University College Cork Distinguished Alumna award for her work. She holds the NSERC/GM Canada Chair for Women in Science and Engineering.


2008/03/14 Speaker: Bill Cook
Affiliation: Georgia Institute of Technology
Title: Numerically Accurate Solutions in Linear and Integer Programming
Abstract: Numerous practical computational problems are formulated as instances of linear or mixed-integer programming. These models are typically described with rational data, while solution software uses floating-point arithmetic and inexact linear algebra to obtain approximate results. Although such solutions are satisfactory in applications, accuracy can be important in many practical and theoretical settings. An important example is the heavy use of linear programming in Hales' proof of the Kepler Conjecture concerning the packing of spheres in three dimensions.
In this talk we treat the problem of finding exact rational solutions to linear and mixed-integer programming models. We describe computational results for benchmark instances, Kepler models, coding-theory bounds, factoring integers, and the traveling salesman problem. This talk is based on joint work with David Applegate, Sanjeeb Dash, Daniel Espinoza, Ricardo Fukasawa, Marcos Goycoolea, and Dan Steffy. Biography of Speaker:

William Cook is the Chandler Family Chair in the School of Industrial and Systems Engineering at Georgia Tech. Cook is the former Editor-in-Chief of Mathematical Programming, Series A (2003--2007) and Mathematical Programming, Series B (1999--2003). He is currently a member of the Editorial Boards of Mathematics of Operations Research, SIAM Journal on Discrete Mathematics, and SIAM Journal on Optimization. Together with David Applegate, Robert Bixby, and Vasek Chvatal, he was awarded the 2000 Beale-Orchard-Hayes Prize by the Mathematical Programming Society and the 2007 Lanchester Prize by INFORMS for joint work on the traveling salesman problem; at an earlier stage, their TSP work was named one of the Top 50 Science Stories of 1992 by Discover Magazine. In 2003, Cook was named the I. E. Block Community Lecturer by the Society for Industrial and Applied Mathematics (SIAM).


2008/03/28 Speaker: Kyros Kutulakos
Affiliation: University of Toronto
Title: Reasoning about Light
Abstract: While research on 3D photography has enjoyed tremendous success in recent years, many everyday objects are still difficult or impossible to scan in 3D. One fundamental stumbling block is that typical algorithms do not consider the effects of light transport---the sequence of bounces, refractions and scattering events that may occur when light interacts with an object. These events are ubiquitous, and dominate the appearance of objects with transparent materials or shiny surfaces.
In this talk, I will present a series of 3D photography algorithms that explicitly reason about how light flows through (or around) real objects. These algorithms rely exclusively on 2D photos and seek to infer the specific optical events that occur from light leaving a source to reaching a specific pixel. I will show that despite the apparent intractability of this endeavor, it has proved quite successful in capturing detailed "3D photos" of many common objects with complex optical properties. Biography of Speaker:

Kyros Kutulakos is an Associate Professor at the University of Toronto where he has been on the faculty since 2001. His research interests are mainly in the area of computer vision, with an emphasis on geometric reconstruction problems. He is the recipient of a Sloan Fellowship, an NSF CAREER award, a PREA award from the government of Ontario, and four best paper prizes (David Marr Prize in 1999, David Marr Prize Honorable Mention in 2005, Honorable Mention at ECCV 2006, and Best Student Paper Award at CVPR 1994). He also served as the Program Co-Chair of the 2003 Computer Vision and Pattern Recognition Conference.


2008/04/03 Speaker: Nicholas Roy
Affiliation: MIT
Title:
Abstract:
2008/04/11 Speaker: Sebastian Fischmeister
Affiliation: University of Waterloo
Title: Separation of Concerns for Non-Functional Properties
Abstract: Separation of concerns is a well established concept in standard software engineering. It means to partition the system into individual parts that overlap as little as possible for specific concerns. Standard software mainly applies this principle to functional properties and behavior. However, embedded and real-time applications have additional concerns tied to non-functional properties such as timeliness, schedulability, and consumption of different resources including computation, communication, and memory. This raises the interesting question whether we can apply the principle of separation of concerns also to non-functional properties and what the consequences would look like. In this talk, I will discuss this question by means of a middleware that separates the concerns of computation, communication, and timing targeted for distributed real-time systems. The discussion ranges from consequences for the system design and system flexibility to the analysis and verification of non-functional and functional properties. Biography of Speaker:

Sebastian Fischmeister is Assistant Professor at the University of Waterloo since February 2008. His research interests lie at the intersection of software technology, distributed systems, and applied formal methods specifically applied to embedded and real-time systems. He received his MASc at the Vienna University of Technology and his PhD degree at the University of Salzburg, Austria. In 2005, he was awarded the Austrian APART stipend for young, excellent researchers and stayed for three years at the University of Pennsylvania as Post-Doctoral Research Associate.