Upcoming Seminars

Mailing list
Register to receive email updates

WebCal feed
Subscribe to seminars in your calendar program

RSS feed
Receive notification of seminar updates in your RSS reader

Seminars

• Software Design in Global Development
  Werner Heijstek (Leiden University)
  Tuesday 23rd March 2010 at 01:10 PM

    Cotton 119, Cotton 119 www.leidenuniv.nl

  Global Software Development (GSD) projects are more often delivered late, over budget and fail considerably more often than `local' software development projects - which have no impressive track record of their own. A rapidly increasing demand for software makes it imperative for software suppliers to participate in this distributed development trend. Initial, cost-related motives no longer drive offshoring as it provides a solution for software organizations that face a shortage of skilled personnel. It is therefore important to find methods to improve the GSD process. A common model for GSD is to create a software design at location A and to have this design implemented at location B. For this purpose, the commonly used Unified Modeling Language (UML) imposes few restrictions. As a result, the build-up of software design documentation and software architecture documentation varies greatly between projects. We investigate design effort, the use of design, design quality, and the use of model-driven methods and GSD process prescriptions to find effective methods of representing and communicating software design. This seminar will give an overview of preliminary results of a set of related studies.

• Bidirectional Transformations: an "Algebra of Programming" approach
  Alcino Cunha (University of Minho, Portugal)
  Thursday 25th March 2010 at 11:00 AM

    , Easterfield EA004

  TBA

• Multi-touch Table User Interfaces for Collaborative Visual Software Analytics
  Craig Anslow
  Tuesday 30th March 2010 at 01:00 PM

    Cotton 119, Cotton 119

  [please note: change in time and location.]

  Most software visualization systems and tools are designed for single users and are bound to the desktop, IDEs, and the web.

  These design decisions do not allow users to collaboratively analyse software or easily interact and navigate software visualizations.

  We are building interactive, collaborative, multi-touch software visualization applications for multi-touch tables. Our user studies will outline the strengths and weaknesses of designing multi-touch software visualization applications and inform users how to collaboratively conduct visual software analytics with multi-touch table user interfaces.

• Organizing Self-Organizing Agile Teams
  Rashina Hoda
  Thursday 1st April 2010 at 11:00 AM

    , Maclaurin 102

  Agile teams are described as "self-organizing". How these teams actually organize themselves in practice, however, is not well understood. Through Grounded Theory research involving 24 Agile practitioners across 14 software organizations in New Zealand and India, we identified six informal roles that team members adopt in order to help their teams self-organize. These roles - Mentor, Co-ordinator, Translator, Champion, Promoter, and Terminator - help teams learn Agile practices, liaise with customers, maintain management support, and remove ineffective team members. Understanding these roles will help software teams become self-organizing, and should guide Agile coaches in working with Agile teams.

• Improving the Performance of Client Honeypot
  Van Lam LE
  Tuesday 20th April 2010 at 01:10 PM

    Cotton 119, Cotton 119

  Malicious web pages have become an emerging security issue nowadays. Its number has been increased significantly and it has raised a special concern from communities. Some approaches have been invented to defense against malicious web pages. Client honeypot is one of the well-known techniques to analyze and study but it is needed to be researched further in order to make it more realistic.

  In this proposal, we present our proposed solution to improve performance of client honeypot in term of detection accuracy and speed. A proposed client honeypot, called Adaptable Client Honeypot, has both high interaction visitor and low interaction visitor to enhance their cooperation in order to classify web pages efficiently. Scoring Mechanism and Classification Mechanism are two main proposed components in our Adaptable Client Honeypot.

  Scoring Mechanism is responsible for finding potential malicious web pages which are then sent to high interaction visitor for further inspections. Because it is the first classifier in our client honeypot system, avoiding missing attack is our main focus. While we would like to have Scoring Mechanism not miss any attack, reducing the number of potential malicious web pages needed to be visited by high interaction visitor is critical. We propose to research on scoring algorithm to score malicious levels of web pages so a threshold can be adjusted in order not to miss attacks but get speed gain. Methodology to develop Scoring Mechanism is going to be proposed in details.

  Classification Mechanism is the final classifier in our proposed client honeypot so its accuracy is the key factor. We propose to research on multiple classifiers to make our Classification Mechanism work more efficiently. Methodology to develop Classification Mechanism is proposed in details.

  We also present the relationship between false positive rate, false negative rate on both Scoring Mechanism and Classification Mechanism, and the overall detection accuracy and speed on our proposed client honeypot. Some main factors are identified as compulsory factors for our implementation of Adaptable Client Honeypot in order to grant detection accuracy and speed.

• Genetic Programming for Classification with Unbalanced Data
  Urvesh Bhowan
  Tuesday 27th April 2010 at 01:10 PM

    Cotton 119, Cotton 119

  In classification problems, learning algorithms can suffer a performance bias when data sets are unbalanced. This happens when at least one class is represented by only a small number of training examples (called the minority class) while the other class(es) make up the majority. In this scenario classifiers can have good majority class accuracy but poor performance on the minority class(es).

  This thesis proposes a Genetic Programming (GP) approach to classification with unbalanced data, aimed at improving the classification ability of both the minority and majority classes in evolved classifiers. We will focus on two major aspects within the GP learning system. The first develops novel new fitness functions in canonical (single-objective) GP to better account for the accuracy of the minority and majority class in fitness. The second uses multi-objective search in GP to simultaneously evolve a Pareto-front of classifiers along the minority and majority class trade-off surface.

  This talk presents the main research objectives and plan for the thesis, an overview of the background and related work in this area, and a summary of the work done so far.

• Whiley: a Language with Extended Static Checkin
  David Pearce
  Tuesday 4th May 2010 at 01:10 PM

    Cotton 119, Cotton 119

  Whiley is a simple programming language aimed primarily at safety-critical applications. The main innovation lies in its ability to perform compile-time checking of pre- and post-conditions. For example, one can specify that an integer parameter must be greater than zero, and the compiler will statically enforce this. The language is split into an imperative outer layer, and a functional inner core. This hybrid design simplifies the problem of statically checking pre- and post-conditions, making it significantly more tractable. This contrasts with efforts to add compile-time constraint checking to more complex languages, such as Java and C#, which have so far been largely unsuccessful.

View seminars in calendar view
Add a seminar

Change your search

You can find historical dates by editing the Date From to before today's date. Leaving the Date From or the Date To fields blank will search for all past or all future dates respectively.