Skip to content. Skip to navigation
McGill Home SOCS Home
Personal tools
You are here: Home Announcements and Events Seminars Seminar History

Seminar Home
Fall 2014 Schedule
Winter 2015 Schedule
Summer 2015 Schedule
Archives

Date
( Fall 2006 )
Category Seminar Info
2006/12/11 CQIL - Cryptography and Quantum Information Place: MC320
Time: 14:00 - 15:00
Speaker: Robin Blume-Kohout
Affiliation: Caltech
Title: All kinds of information
Abstract: Information is physical; it's always about something. Specifically, it always describes some set of observables. I'll present a proof that the kinds of information that can exist self-consistently correspond to associative matrix algebras. Both quantum and classical information fit into this general framework of information-preserving structures. This yields an algorithm for finding all the noiseless subsystems of a quantum process.
2006/12/06 Bioinformatics Place: Duff Ampitheatre 1
Time: 14:30 - 15:30
Speaker: Reza Salavati
Affiliation: McGill University, Institute of Parasitology
Title: TBA
Abstract:
2006/12/06 Algorithms Place: MC320
Time: 16:00 - 17:00
Speaker: Pat Morin
Affiliation: School of Computer Science, Carleton University
Title: Distribution-Sensitive Point Location in Convex Subdivisions
Abstract: We consider the problem of preprocessing a convex polygon, P, and a probability measure, D, over the plane so that, for a query point p selected according to D we can quickly test if P contains p. Despite the long history of planar point location data structures in computational geometry, including results on distribution-sensitive planar point location, it seems that this simple problem has not been studied before. In this talk we will describe a hierchical triangulation T=T(P,D) of the plane that simultaneously serves two purposes: 1. Using T we obtain a data structure for testing if a query point p is contained in P. 2. Using T we obtain a lower bound on the expected cost of any linear decision tree that tests if a point p drawn from D is contained in P. The expected query time of the data structure (1) matches the lower bound (2) to within a constant factor, thereby proving that the data structure is optimal. An easy consequence of these results is an optimal data structure for planar point location in convex subdivisions when the queries come from a known distribution.
2006/12/06 CQIL - Cryptography and Quantum Information Place: Postponed
Time: -
Speaker: Omar Khalid
Affiliation: McGill University
Title: On fault-tolerant quantum computation
Abstract: Biography of Speaker:


2006/12/01 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: Haiying Xu
Affiliation: McGill University
Title: Method Purity Analysis and Memoization for Java Programs
Abstract: he knowledge of method purity has many important applications. And depend on different motivation, there are some different criteria for pure methods. Basically, a pure method can not mutate it's pre-state, that is cannot mutate the objects exists before method invocation. In our research, we perform both static and runtime purity analysis. The analysis at runtime is based on three different purity criteria. After make a comparison between those analysis. We begin to combine one of those analysis with memoization to reduce the execution time, which can find much more pure methods than others. The result shows that there is almost no overhead for memoization, all most all the overhead is come from runtime purity analysis.
2006/11/24 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: Eric Bodden
Affiliation: McGill University
Title: A staged static program analysis to reduce the performance overhead of runtime monitoring
Abstract: A program is said to work correctly if it adheres to its specification. In order to verify a program for correctness, runtime monitoring uses specifications that describe sequences of events which must not occur during program execution. This specification is checked by a monitor as the program runs. We were faced with two main problems when trying to implement runtime monitoring efficiently: (1) implementing the induced finite state monitor efficiently and (2) reducing the number of instrumentation points which make calls to this monitor. This paper addresses the second point using static program analysis. Our current work focuses on trace matches which allow programmers to specify sequences of events via regular expressions with free variables. We present a staged analysis to reduce the performance overhead of tracematches at runtime, where each stage has different precision/cost tradeoffs and builds up on the results of the former stages. The first is very cheap and high level. Then, in the second stage, a flow-insensitive whole-program analysis is applied. The final stage is flow-sensitive, taking into consideration the order in which the events occur. We implemented all three analyses in the AspectBench Compiler and experimented with a collection of tracematches applied to the DaCapo benchmark suite.
2006/11/22 Bioinformatics Place: Duff Ampitheatre 1
Time: 14:30 - 15:30
Speaker: Ajjamada Kushalappa
Affiliation: McGill University, Dept. of Plant Sciences
Title: TBA
Abstract:
2006/11/22 Vision, Graphics, and Robotics Place: ENGMC 437
Time: 11:00 - 12:00
Speaker: Chris Pal
Affiliation: University of Massachussetts
Title: Representations, Algorithms and Interactive Visual Media
Abstract: Standard video is a simple stack of rectangular images of fixed resolution, displayed linearly in time. Can we create a representation of a scene that places dynamic, high resolution content in the directions we are most likely to look? I tell you how to do this. I start with a top down view of how to decompose a complex scene using probabilistic generative models. I contrast the computational efficiency and representational power of different models. I then focus on the goal of efficiently constructing a high quality representation and re-examine the tasks involved. I outline the critical computations required for registering high-definition quality video frames into huge, high quality panoramic representations. I show how performing this task effectively and efficiently then facilitates the construction of rich, animated and interactive representations. Finally, I show how effective video registration also empowers an interactive system for real-time panoramic photography.
2006/11/17 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: Guven Bolukbasi
Affiliation: McGill University
Title: Extending AspectOPTIMA: The Design of Transactional Life Cycle with Reusable Aspects
Abstract: The AspectOPTIMA project aims at creating an aspect-oriented framework that provides run-time support for transactions. Decomposing the ACID (atomicity, consistency, isolation, durability) properties into ten well-defined reusable aspects may be considered as the first step in the project. However, an extensive analysis of this decomposition revealed a cross-cutting concern among the aspects. As a follow-up work, the current research tries to refactor the existing framework and also extend it by adding the transactional life cycle management. The presentation will discuss the details of this refactoring & integration process by focusing on elements such as transaction boundaries/zones, states of a transaction, nesting and member synchronization.
2006/11/15 Bioinformatics Place: Duff Ampitheatre 1
Time: 14:30 - 15:30
Speaker: Benoit Coulombe
Affiliation: IRCM
Title: A panorama of the protein complexes formed by the general transcription machinery in mammals
Abstract: We have performed a survey of protein complexes containing transcription and RNA processing factors in the soluble compartment of mammalian cells using protein affinity purification coupled to mass spectrometry. High confidence interactions were selected computationally. Thirty-four tagged polypeptides yielded a network of 842 interactions. Remarkably, the network is significantly enriched in proteins known to regulate the formation of protein complexes, uncovering a novel regulatory mechanism that targets the transcription machinery. The network also contains previously-uncharacterized proteins to which we have started inferring functions. The RNA polymerase II (RNAPII)-associated proteins (RPAPs) are physically and functionally associated with RNAPII, forming an interface between the enzyme and chaperone and scaffolding proteins. The newly-discovered methylphosphate cap synthase MePCS is part of a 7SK/U6 snRNP complex containing RNA processing and transcription factors, including the elongation factor P-TEFb. Our results define the first high-density interaction network involving the cellular machinery that interprets mammalian genomes. Biography of Speaker:


2006/11/15 Algorithms Place: MC103
Time: 16:00 - 17:00
Speaker: Mark Grundland
Affiliation: University of Cambridge
Title: Improving Image Quality in Image Compositing
Abstract: In image and video editing, image compositing is the process of combining pictures, commonly know as photomontage. A key technical challenge is to producecomposites that preserve the visual quality of the components used to create them. Linear interpolation is the standard method of image blending used in image compositing. As an averaging operation, linear interpolation acts to reduce variation. As a result, its output can look less appealing than its input. Composite images often leave something to be desired. When creating layered visual effects, compositing artists may need to manually compensate for faded contrast, dull colors, lost details, or lack of emphasis. To address theselong standing shortcomings of image blending, we redefine the fundamental operation of linear interpolation, reformulating its weights, its operators, andits results. We aim to give users visually meaningful, high level control over the appearanceof composite images. We propose four novel image blending operators that are designed to maintain key visual characteristics of their inputs: contrast, color, detail, and salience. Our contrast preserving technique applies a linear color mapping to recover the contrast lost due to linear interpolation. Our color preserving technique extends homomorphic image processing by establishing an isomorphism between the image colors and the real numbers, allowing any mathematical operation defined on real numbers to be applied to colors without losing its algebraic properties or mapping colors out of gamut. Our detail preserving technique relies on the signed weighted power mean to emphasize variation over uniformity combining image features at different levels of detail. Finally, our salience preserving technique retains the most informative regions of the input images by balancing their relative opacity with their relative saliency. Applying our approach, casual users may benefit from image compositing tools that do not require highly accurate image mattes to create compelling composites, while expert users may take advantage of contrast enhancement and color correction tools to ensure sharp contrast and vivid colors. This seminar presents the work of Mark Grundland, Rahul Vohra, Gareth P. Williams, and Neil A. Dodgson at the University of Cambridge, which was recentlypublished as "Cross Dissolve Without Cross Fade: Preserving Contrast, Color and Salience in Image Compositing" at EUROGRAPHICS and "Nonlinear Multiresolution Image Blending" at ICCVG.
2006/11/08 Bioinformatics Place: Duff Ampitheatre 1
Time: 14:30 - 15:30
Speaker: David Juncker
Affiliation: McGill University
Title: TBA
Abstract: Biography of Speaker:


2006/11/03 CQIL - Cryptography and Quantum Information Place: Rutherford 103
Time: 13:30 - 14:30
Speaker: Stephen Hsu
Affiliation: University of Oregon
Area: Quantum Information and Theoretical Physics
Title: Entanglement entropy, black holes and holography
Abstract: After an introduction to black hole entropy and holography, we describe results related to the entanglement entropy of black holes. We observe that the entanglement entropy resulting from tracing over a subregion of an initially pure state can grow faster than the surface area of the subregion (indeed, proportional to the volume), in contrast to examples studied previously. The pure states with this property have long-range correlations between interior and exterior modes and are constructed by purification of the desired density matrix. We show that imposing a no-gravitational collapse condition on the pure state is sufficient to exclude faster than area law entropy scaling. This observation leads to an interpretation of holography as an upper bound on the realizable entropy (entanglement or von Neumann) of a region, rather than on the dimension of its Hilbert space. 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.


2006/11/01 Algorithms Place: MC320
Time: 16:00 - 17:00
Speaker: Olivier Devillers
Affiliation: INRIA Sophia Antipolis
Title: Succinct Representations of Triangulations and Planar Maps
Abstract: In this talk, we address the problem of representing the connectivity information of geometric objects using as little memory as possible. As opposed to raw compression issues, the focus is here on designing data structures that preserve the possibility of answering incidence queries in constant time. We propose in particular the first optimal representation for 3-connected planar graphs and triangulations, which are the most standard classes of graphs underlying meshes with spherical topology. Optimal means that these representations asymptotically match the respective entropy of the two classes, namely 2 bits per edge for 3-c planar graphs, and 1.62 bits per triangle or equivalently 3.24 bits per vertex for triangulations. This work presents a rather theoretical point of view, in conclusion we will discuss the possibility of more practical versions. This is a joint work with Luca Castelli Aleardi, Abdelkrim Mebarki and Gilles Schaeffer (appeared WADS 2005, CCCG 2005, SoCG 2006).
2006/11/01 Bioinformatics Place: Duff Ampitheatre 1
Time: 14:30 - 15:30
Speaker: Shoshana Wodak
Affiliation: University of Toronto
Title: TBA
Abstract:
2006/10/31 Software Engineering Place: MAAS 328
Time: 9:00a - 9:55a
Speaker: Silvia Breu
Affiliation: University of Cambridge
Area: Software Evolution
Title: Mining Aspects from Version History
Abstract: Over the past few years, aspect mining has become an active area of research. It tries to identify cross-cutting concerns in a program to help migrating it to a better design, maybe even to an aspect-oriented design. Previous approaches to aspect mining applied static or dynamic program analysis techniques to a single version of a system. As a result, aspect mining has difficulties to scale to large systems because it depends for good coverage either on many test cases (dynamic approaches), or on difficult incremental analysis (static approaches).

In this informal talk, I will present a novel approach that leverages a system's version history to mine aspect candidates. By analysing where developers add code to a program, our history-based aspect mining (HAM) identifies and ranks cross-cutting concerns. This process is independent from the total size of a system as we analyse changes from one version to the next. Moreover, since HAM is static, it does not rely on test cases but guarantees complete coverage. As a result, we are able to mine industrial-sized software systems such as Eclipse. Furthermore, we believe that history-based aspect mining will provide a completely new view on the evolution of cross-cutting concerns themselves.

In addition, I will give a short outlook for discussion on our current work that focusses on the open question of how to combine mined simple aspect candidates into complex ones that then represent full aspects. As the straightforward combination of those candidates that are found in identical locations in the code only works for simple cases, we employ formal concept analysis to CVS transactions for the general case. Biography of Speaker:

Silvia Breu graduated with a diploma degree in Computer Science, Math, and Applied Foreign Languages from the University of Passau in 2004, having done part of her studies at the University of Edinburgh. During the summer of 2004 she was a research intern at the NASA Ames Research Center within the Automated Software Engineering Group. With her work on dynamic aspect mining, she won the Ernst Denert Software Engineering Award in 2004, which is an award for the best MSc thesis in Germany. Before starting her PhD in 2006 in the Computer Laboratory at the University of Cambridge, supported by a scholarship from the Bill & Melinda Gates Foundation and a Neil Wiseman Memorial award, Silvia worked as a researcher at the Software Engineering Chair at Saarland University. She is a member of Newnham College. Her research interests cover program analysis (aspect mining, program slicing), software evolution, as well as programming languages, particularly functional languages such as Haskell.


2006/10/27 Software Engineering Place: McConnel 103
Time: 14:00 - 15:00
Speaker: Michael Batchelder
Affiliation: McGill University
Title: Java Bytecode Obfuscation
Abstract: Java, due to its portable design and open specifications, is a language vulnerable to intellectual property attacks - specifically reverse-engineering through automatic decompiler tools. While this decade has seen a marked rise in protection schemes such as encryption and digital rights management, obfuscation can still serve as a first-defense and has not yet been fully explored. In this talk three categories of obfuscations are discussed, low-level operations, high-level structure, and gap-exploits (obfuscations that exploit the gap between the high-level Java source code language and its low-level cousin Bytecode). A number of new and never before seen obfuscations will be presented.
2006/10/25 Algorithms Place: MC320
Time: 16:00 - 17:00
Speaker: Christophe Paul
Affiliation: Laboratoire d'Informatique de Robotique et de Microelectronique de Montpellier
Title: Algorithmic Aspects of Modular Decomposition
Abstract:
2006/10/25 Bioinformatics Place: Duff Ampitheatre 1
Time: 14:30 - 15:30
Speaker: David Stephens
Affiliation: McGill University, Dept. of Mathematics and Statistics
Title: TBA
Abstract: Biography of Speaker:


2006/10/20 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: Richard Halpert
Affiliation: McGill University - Sable group
Title: Lock Allocation for Pessimistic Transactional Programming
Abstract: As multiple-processor systems become more prevalent on the desktop, it is becoming more important for consumer-grade multithreaded programs to be correctly synchronized. Unfortunately, correct synchronization is difficult to achieve using locks, synchronized regions, and other existing techniques. A multithreaded programming technique called Transactional Programming promises to ease the difficulties of correct synchronization. In this talk, I will present a technique to compile transactional Java programs to Java bytecode that uses existing synchronization primitives to ensure correct synchronization. Unlike optimistic (runtime) implementations of transactions for Java, this implementation does not require any hardware, JVM, or library changes.
2006/10/18 Algorithms Place: MC320
Time: 16:00 - 17:00
Speaker: Katsuhisa Yamanaka
Affiliation: Gunma University, Japan
Area: graph theory
Title: A Compact Encoding of Rectangular Drawings with Efficient Query Support
Abstract: A rectangular drawing is a plane drawing in which every face is a rectangle. In our research, we give a simple encoding scheme for rectangular drawings. The encoding scheme needs 5m/3 + o(n) bits for each n-vertex rectangular drawing R, where m is the number of edges of R, and supports a rich set of queries, including adjacency and degree queries, in constant time.
2006/10/13 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: Barthelemy Dagenais
Affiliation: McGill University
Title: Guidance through Active Concerns
Abstract: Producing usable documentation has always been a tedious task, and even communicating important knowledge about a system among collaborators is difficult. This talk describes an approach to creating documentation in the form of guides, which encapsulate passive information about important tasks along with active steps to be followed. The approach is concern-based, and introduces active steps into traditionally passive concerns. A developer can begin by creating a concern that identifies elements of importance in the context of a task, which, we believe, is easier and more natural than trying to formulate a process up front. S/he can then easily create a guide to the task based on this concern, and export it. Other developers can follow the guide, and, as they do so, their results are recorded as examples for future reference. As an early step towards validation, we created a guide for the complex task of creating an Eclipse editor.
2006/10/12 CQIL - Cryptography and Quantum Information Place: Redpath Museum Audit
Time: 18:00 - 19:00
Speaker: Claude Crepeau
Affiliation: McGill University
Area: The Cutting Edge: Royal Society Lectures in Science
Title: Alice and Bob's Strange Adventures in Quantumland
Abstract: Computer science is facing an unavoidable frontier: as we keep building smaller and smaller computer components, within a few decades, we will hit the scale where the rules of nature are quantum mechanical instead of classical. This could be considered as a drawback, but as it will be explained in this talk, it appears to be a "better" world: A world where information processing can solve certain tasks faster, more completely or achieve situations impossible in our classical world. The big question is: "Will we be able to tame the quantum world and build this amazing computer?" Follow Alice and Bob through a journey that will explore the new possibilities of the quantum computer. No background on quantum physics is expected.
2006/10/11 Algorithms Place: MC320
Time: 16:00 - 17:00
Speaker: Godfried Toussaint
Affiliation: McGill University
Title: Computational Geometric Aspects of Musical Rhythm
Abstract:
2006/10/06 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: Michael Batchelder
Affiliation: McGill University - Sable group
Title: Java Bytecode Obfuscation
Abstract: Java, due to its portable design and open specifications, is a language vulnerable to intellectual property attacks - specifically reverse-engineering through automatic decompiler tools. While this decade has seen a marked rise in protection schemes such as encryption and digital rights management, obfuscation can still serve as a first-defense and has not yet been fully explored. In this talk three categories of obfuscations are discussed, low-level operations, high-level structure, and gap-exploits (obfuscations that exploit the gap between the high-level Java source code language and its low-level cousin Bytecode). A number of new and never before seen obfuscations will be presented.
2006/09/29 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: Jörg Kienzle
Affiliation: McGill University
Title: Model-Driven Assessment of Use Cases for Dependable Systems
Abstract: Complex real-time systems need to address dependability requirements early on in the development process. This talk presents a model-based approach that allows developers to analyze the dependability of use cases and to discover more reliable and safe ways of designing the interactions with the system and the environment. We use a probabilistic extension of statecharts to model the system requirements. The model is then evaluated analytically based on the success and failure probabilities of events. The analysis may lead to further refinement of the use cases by introducing detection and recovery measures to ensure dependable system interaction. A visual modelling environment for our extended statecharts formalism supporting automatic probability analysis has been implemented in AToM3, A Tool for Multi-formalism and Meta-Modelling. Our approach is illustrated with an elevator control system case study.
2006/09/27 Algorithms Place: MC320
Time: 16:00 - 17:00
Speaker: Steph Durocher
Affiliation: McGill University
Title: Geometric Facility Location under Continuous Motion
Abstract:
2006/09/22 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: Martin Robillard
Affiliation: McGill University
Title: Tracking Concerns in Evolving Source Code: An Empirical Study
Abstract:
2006/09/22 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: Jean-Sebastien Boulanger
Affiliation: McGill University
Title: Managing Concern Interfaces
Abstract:
2006/09/20 Bioinformatics Place: Duff Ampitheatre 1
Time: 14:30 - 15:30
Speaker: Mathieu Blanchette
Affiliation: McGill University
Title: Whole-genome comparative and regulatory genomics
Abstract: This talk will describe how a whole-genome computational prediction and analysis of human regulatory regions can yield important insights into gene regulation, and how genome evolution, and in particular computationally reconstructed ancestral DNA sequences, can help in this process. I will first describe a approach to the detection of cis-regulatory modules that exploits both inter-species comparison and binding site clustering. The analysis of the ~120,000 modules identified by this algorithm reveals a number of interesting observations regarding the overall distribution properties of the modules, but also regarding the properties of the individual transcription factors predicted to bind them. These properties include association to particular expression patterns or function, co-occurrences of binding sites for pairs of transcription factors, and broad regulatory network properties. In the second part of the talk, I will briefly introduce a joint project with Dr. David Haussler and Dr. Webb Miller, aiming to reconstruct the complete genome of ancestral mammals. I will focus on how this ancestral sequence information can help our study of the evolution of regulatory mechanisms in mammals, and how these sequences can be used to predict human regulatory regions more accurately.
2006/09/15 Software Engineering Place: McConnell 103
Time: 14:00 - 15:00
Speaker: SE & Sable Group
Affiliation: McGill University
Title: Welcome and Introductions
Abstract: The goal of this seminar will be to welcome the new members of the various research groups, do a round of introductions, and plan the schedule for the term.
2006/09/15 CQIL - Cryptography and Quantum Information Place: Rutherford 112
Time: 15:30 - 16:30
Speaker: Patrick Hayden
Affiliation: McGill University, School of Computer Science
Area: Physics dept. colloquium
Title: The power of forgetting
Abstract: Thermodynamics places surprisingly few fundamental constraints on information processing. In fact, most people would argue that it imposes only one, known as Landauer's Principle: a process erasing one bit of information must release an amount kT ln 2 of heat. It is this simple observation that finally led to the exorcism of Maxwell's Demon from statistical mechanics, more than a century after he first appeared. Ignoring the lesson implicit in this early advance, however, quantum information theorists have been surprisingly slow to embrace erasure as a fundamental primitive. Over the past year, however, it has become clear that a detailed understanding of how difficult it is to erase correlations leads to a nearly complete synthesis and simplification of the known results of asymptotic quantum information theory. As it turns out, nearly all the tasks of interest, from distilling high-quality entanglement to sending quantum data through a noisy medium to many receivers, can be understood as variants of erasure. I'll sketch the main ideas behind these discoveries and end with some speculations on what lessons the new picture might have for understanding information loss in real physical systems.
2006/09/13 General Place: MC320
Time: 16:00 - 16:30
Speaker: Ethan Kim
Affiliation: Computer Science/Discrete Math
Area: Computer Science
Title: On Bus Graph Realizability
Abstract: In this talk, we consider the following graph embedding problem: Given a bipartite graph G = (V_1; V_2; E), where the maximum degree of vertices in V_2 is 4, can G be embedded on a two dimensional grid such that each vertex in V_1 is drawn as a line segment along a grid line, each vertex in V_2 is drawn as a point at a grid point, and each edge e = (u,v) for some u \in V_1 and v \in V_2 is drawn as a line segment connecting u and v, perpendicular to the line segment for u? We give a proof sketch that this problem is NP-complete, and discuss how the proof techniques can be used to show the hardness of several other related problems.
2006/09/13 Bioinformatics Place: Duff Ampitheatre 1
Time: 14:30 - 15:30
Speaker: Ted Perkins
Affiliation: McGill University
Title: Transcriptional Regulation: Character and Capabilities
Abstract: Transcriptional regulation, discovered roughly a half-century ago by Jacob and Monod, is one of the major means by which the concentrations of gene products are controlled. However, the relationships between transcription factor concentrations and the expression level or transcription rate of a gene are very complex. As a result, there is a lack of consensus on how to express these relationships---that is, on what kinds of models capture the essential features of transcriptional regulation without eliminating important details. This is a problem, of course, for those interested in modeling transcriptional regulation. But it is also a problem for those who want to understand and convey the behavior of particular genes, as well as those who want to engineer genes. This talk will be a three-part story on the character of transcriptional regulation. In the first part, I will survey some of the important historical milestones in our understanding of transcriptional regulation, beginning with the seminal work of Jacob and Monod. In the second part, I will describe lessons learned from detailed modeling of "high-fidelity" expression data sets, including some of my own work on the segmentation network of Drosophila melanogaster. In the final part, I will describe a thought experiment: Given what we know about transcriptional regulation, what kinds of relationships between transcription factors and target gene are possible?
2006/08/24 Software Engineering Place: MC103
Time: 10:30 - 11:30
Speaker: Tom Zimmermann
Affiliation: Saarland University, Germany
Title: Mining Usage Pattern from Version Archives
Abstract: When developers change code they insert new method calls.  Typically, method calls that are inserted together, are related to each other.  In this talk, I will present the DynaMine tool that obtains this co-addition relationship from version archives in order to identify usage patterns.  Such patterns describe how methods should be called, for instance a call to addWidget() should be followed by removeWidget().  Besides simple pairs, DynaMine also discovers state machines.  Potential patterns are then validated dynamically and violations are reported to the user.  DynaMine has been evaluated with the history of open-source projects and scales up to industrial-sized projects like Eclipse. Biography of Speaker:

Tom Zimmermann received his diploma degree in computer science from the University of Passau in 2004.  He is currently a PhD student at the Saarland University in Saarbrücken, Germany, supported by a scholarship from the DFG research training group on Quality Guarantees for Computer Systems.  Tom built the eROSE tool, which helps developers navigating through source code.  eROSE leverages the change history of projects and learns recommendations such as "Programmers who changed function f() also changed function g()".  Besides supporting developers, Tom's research interests are in software evolution, program analysis, and data mining.


2006/08/17 Software Engineering Place: MC103
Time: 10:30 - 11:30
Speaker: Isaac Yuen
Affiliation: Software Evolution Research Group
Title: Assembling the scattered code: An overview of aspect-oriented refactoring techniques and challenges
Abstract: Aspect-oriented programming (AOP) enables a modular implementation of crosscutting concerns in a system. Aspect-oriented refactoring is the mechanism that transforms an existing object-oriented system into its AOP equivalent. Nevertheless, there are two questions a developer must answer when he decides to use AO refactorings: what constitute the refactorable crosscutting concerns in the system, and what techniques are required to extract them into aspects. In this presentation, I will describe the applicability of AO refactoring using a case study (jEdit), and introduce some common AO refactoring tech- niques. I will also describe two different approaches that automate the refactoring process and their shortcomings. I will conclude the presentation by describing the current research in this area, and the development of refactoring assessment tool. Biography of Speaker: