Design and Evaluation of a Lightweight Persistent Data Store for High-Frequent Concurrent Read and Write Accesses

This research project will systematically compare different design alternatives for a data store module that is tuned for concurrent accesses by multiple independent threads in a program. The particular access pattern is modelled after the specifics of automated securities trading programs implemented in Java. Based on a systematic evaluation, a prototype implementation will be developed that will allow X3 Trading to dramatically improve the efficiency of business application development.

Management and mining of big energy sector data

With advances in techniques, high volumes of valuable data are generated in many domains (e.g., energy sector) at a rapid rate. Consequently, a scalable and flexible system for efficient storage and fast management of these distributed data is needed. In this proposed research project, we plan to design and implement a cloud-based data storage & management system that is flexible, scalable and fast to handle distributed data in a parallel fashion for the partner organization.

Automated optimization of robotic tasks and transitions using graph-based approaches

In recent years, machining with robots has become a trend in the manufacturing industry. The concept offers an economical solution for medium to low accuracy machining applications. However, due to the complexity of the robot kinematics, planning for these paths is challenging. Jabez Technologies has developed a semi-graphical approach that can program large robot-paths. This approach has been very well received by the industry and has proven to be extremely robust in practice. However, this approach is semi-automatic and cannot work without user input.

Big Data Cleaning

Poor data quality is a barrier to effective, high-quality decision-making based on data. Declarative data cleaning has emerged as an effective tool for both assessing and improving the quality of data. In this work, we will address some important challenges in applying declarative data cleaning to big data, challenges that arise due to the scale, complexity, and massive heterogeneity of such data. First, we will investigate the use of domain ontologies to enhance declarative data cleaning. Second, given the dynamic nature of big data, we will develop new continuous data cleaning methods.

Long-dated foreign exchange interest rate hybrid financial derivatives: modeling,calibration, pricing, and risk-management

The proposed project addresses three major challenges in modeling long-dated (maturities of 30 years or more) foreign exchange (FX) interest rate (IR) hybrid derivatives, namely

(i) the strong sensitivity of the products to the skew of the FX volatility smiles,

(ii) their very long maturities, and

(iii) popular embedded exotic features, which provide possibilities of early termination of the products.

Increasing and Automating Adaptivity of LogicBlox Datalog Platform

Declarative programming techniques have been used to make programming available to common users through introducing query languages such as SQL and spreadsheet programs such as Excel. The proposed industrial partner, LogicBlox, has developed software that enables highly-complex data analysis through the flexible and familiar form of spreadsheet computing. The main goal of this proposal is to increase and automate adaptivity of LogicBlox platform, and to evaluate proposed methodology through applications in local industry.

Managing Knowledge Discovery and Data-Mining Through Visual Analytics, Concept Maps and Ontologies

This research project aims to prototype and evaluate a computer system supporting the creation of data-driven domain ontologies.
The first phase consists in creating a working prototype of a computer application supporting:
1) Preliminary research;
2) Creation of an interactive and visual computer interface allowing the visual concept mapping of a domain, its entities; define relationships between those entities and define attributes and parameters of those entities and relationships;

Development of a smartphone-based sensor system to assess everyday movement, speech, and sleep for stroke recovery Year One

Following a stroke, the rehabilitation gains achieved in intensive therapy are often lost without sustained follow-up. The long-term objective of the project is to extend the reach of the health team (physicians, therapi sts, caregivers) following hospital discharge to continue therapy using mobile devices. As a first step, the objectives of the planned project are to:

1) develop clinically meaningful metrics related to the quantity and quality of everyday movement, speech, and sleep from body-worn sensor data,

2) to design and fabricate system prototypes for testing, and

Advanced Adaptivity and Personalization in Learning Systems through Collaborative Recommendations Year Two

Learning Systems are among the most popular e-learning tools in today’s education and training. Most e-Learning systems do not take into account individual aspects of learners (e.g., their goal, experiences, existing knowledge, learning style etc.).The primary goal of the  proposed research is to offer rich adaptivity by combining information from a learner’s profile (e.g. levels, goals, learning style, cognitive abilities etc) with the information from other learners sharing common interests.

Social Lead Identification Year Two

Millions of people post information on social media sites about their interests, preferences, opinions etc. on a daily basis. LeadSift mines this data stream in real-time to generate incredibly accurate and targeted sales leads. Given the short text and ambiguity around a social post, it gets very difficult to accurately identify intent. This project will explore different Natural Language Processing techniques to analyze the inherent semantic structure of social posts. Due the real-time nature of the social data, the algorithms need to be extremely efficient and scalable.

Pages