Accelerating Learning Process for Medical Image Segmentation using Opposition-based Learning

In this project we will translate and test artificial intelligence techniques to segment medical images with the capability to perform more accurately than conventional methods. The technique is expected to largely eliminate tedious manual delineation of suspicious objects in medical images or, at least, significantly reduce the intensity and frequency of manual modifications traditionally preformed by radiologists.

Recovery for Web Service Conversations

Increasingly, global business relies on the exchange of information between web services. It is crucial that these services correctly exchange messages. Currently, unexpected errors are fixed in an ad-hoc manner, like sanitizing the database and restarting the application.

Large Inbox Voice Interaction Design

A traditional linear speech interface presents significant usability challenges when applied to large databases, like an inbox full of email. This presents an Interaction Design problem to overcome the inherent shortcomings of traditional voice interfaces for this application. The study will examine how users process their email using iLane (an in-vehicle voice interactive platform that seamlessly interfaces with existing smartphones), then design alternatives, refine, and test the selected alternative.

Analysing the Information Seeking, Use and Sharing Practices of Newcomers to Canada

The proposed study explores the information practices of new immigrants in large, urban cities in Canada with respect to their choices regarding mobile phone services. The research goals are to (1) assess how mobile phones fit within the information practices of newcomers (2) understand when and where newcomers get information about mobile phone services in Canada, and(3) identify patterns of use and perceptions of service about mobile technologies among newcomers.

Integration and Assessment of Compiled Parallel Bit Stream Technology in High Performance Text Processing Applications

International Characters, Inc., based on research at Simon Fraser University, has developed high-performance software techniques for reading text files in various formats into databases. However, programming using those techniques can be difficult and error-prone. New research has shown that this form of programming can be automated to a considerable extent. This project is to evaluate the automatically-generated software and determine how easily it can be integrated into important commercial applications.

Efficient Modeling of Large-Scale LIDAR Date

Softree Technical Systems Inc has been providing innovative software solutions for surveying, mapping and engineering for more than 15 years. Softree's main mission is to provide high quality software applications in land development, civil engineering and the naturalresource industries. Recently, the company has developed a robust software to process the terrain data gathered from a LIDAR system. The main objective of the application is to provide the user with the ability to draw, label, manipulate, and extract some features of the terrain.

Access Patterns in Social Games: Detecting Cheaters in a Virutal Economy

This research internship is centered on the discovery of specific behavioral trends among vast collections of user data; more specifically,the identification of cheaters among a community of social online gamers. The current approach to fraud detection, taken by Backstage Technologies Incorporated, is highly manual and heuristic. We plan to experiment with modern data mining and machine learning techniques to more accurately identify cheating trends within a dataset. Further, we plan to implement a solution based on recent literature regarding code modularity and flexible design.

Activities of Daily Living (ADL) Recognition

The project shall explore the cutting-edge research and technologies in monitoring daily activities using a set of sensors deployed in the house. The objective of the research is to provide a feasible solution for improving care for patients with chronic disease, while significantly reducing the healthcare cost. We will conduct a comparative study on the existing models for characterizing human activities, and algorithms for activity recognition.

Design and Development of a Usable User Interface for a Search Engine in the Oil and Gas Domain

Currently, in the oil and gas industry, document handling tasks are performed manually. Intellog is developing a search engine for handling the documents to various users in this domain. Due to complexity of the search task and variety of the users, development of a usable user interface for the search engine is challenging. In this project usability engineering methods will be used to develop a user interface, which will be integrated into the intellog's software.

Software Tools for the Compression of Sets of Satellite Images

In conjunction with Iunctus Geomatics Corp, we propose to develop a storage system for collected satellite image sets. The properties of typical sets of satellite images will be used to reduce the overall storage requirement for the image sets. This is expected to lead to reduced hardware requirement for storage, as well as a reduction in transmission time for image sets. Software tools to manage the image sets will be developed.