Learning representations through stochastic gradient descent by minimizing the cross-validation error

Representations are fundamental to Artificial Intelligence. Typically, the performance of a learning system depends on its data representation. These data representations are usually hand-engineered based on some prior domain knowledge regarding the task. More recently, the trend is to learn these representations through deep neural networks as these can produce significant performance improvements over hand-engineered data representations. Learning representations reduces the human labour involved in any system design, and this allows in scaling of a learning system for difficult problems.

Learning tools to predict treatment responses for schizophrenia from neuroimaging data

Schizophrenia is a chronic mental disorder associated with a significant health, social and financial burden, not only for patients but also for their families, and society. However, the current treatment methods have been only partially successful, mainly due to the inter-individual differences between patients, which means that a treatment that is successful for one patient, might not work for another.

Searchable Social and Environmental Impact Measurement Database

Stakeholders from all business sectors are increasingly looking to businesses to address pressing social and/or environmental issues. Co-operatives are facing the same challenges, and must also use non-financial indicators to demonstrate their co-operative difference.
The purpose of this project is to develop a web-based searchable database of existing tools and frameworks to measure social and environmental performance of business and enterprises.

Enhancing Recommendation Engine for Open Source Software with Community Structures and Copulas

In the constantly expanding world of open source software and services, developers find it increasingly difficult to choose open source that is compliant, secure and reliable. There are millions of open source software and services publicly available today, and compliance, security and quality related information is extremely difficult for developers to find, making mindful selection of open source an onerous process.

Data supervision and security in large data repositories

Ensuring data security in large data repositories is a challenging task as the volume and the nature of the data to secure constantly evolves. Large repositories are mostly composed of documents expressed in natural language and as a result they are a rich source of information. Given the importance of personal data protection, this proposal explores new methods to mine networks of communications between users and detect improper dissemination of sensitive information.

An Integrated Mobile Communication Environment for Healthcare Professionals and Patients: Analysis, Prediction and Recommendation

Think Tank Innovations Ltd is a company active in the healthcare domain. The company is interested in expanding their current working system Sharesmart application by benefiting from advanced technology to develop an integrated environment for healthcare solutions in order to better service a wider community locally and globally. Students to be involved in this project will build a data repository to host data to be collected, cleaned, built, integrated and processed for knowledge discovery which will guide more focused decision making.

Advanced Analytics for Credit Unions

The worldwide data explosion would emphasize on importance of knowledge discovery from massive, heterogeneous, and dynamic volumes of information (Big Data). Similar to other industrial organizations, Credit union industry deals with massive amounts of structured (e.g. customer demographics and transactional data) and unstructured (e.g. email, social media data, comments) data which they have not utilised well to be able to proactively offer their products and services according to their customer needs.

Applying Deep Learning to Optimize 3D Pose Estimation from Monocular Video

REP is an athlete development platform for building better, and healthier athletes. Inside the REP platform are computer algorithms that can “see” how people move, and the accurately estimate how they are moving in three dimensions. The REP platform can then compare models of how you move, to models of how experts move. This comparison gives us rich information that people can use to improve their form. However, generating the expert models is quite hard, and it’s not always easy to understand how to actually compare users and experts.

Leisure Access Victoria: Recreation Accessibility Website, Apps and Tools

Recreation Integration Victoria and the School of Public Health and Social Policy (at UVic.) will address critical issues around the health, fitness and social integration of persons with disabilities in the Victoria Capital Regional District (CRD). Our goal is to promote and facilitate increased fitness, physical activity and healthy living across the entire disability spectrum.

Security assessment of mobile financial services

Information technology rapid evolution was always closely followed by sophistication of malware. And with ubiquitous shift to mobile platforms, rise of mobile malware and in particular banking malware came as no surprise. In general, any financial operation on a mobile platform potentially exposes a user to the variety of threats including data leakage, theft and financial loss. Driven by financial profits, banking malware leverages user's cluelessness, openness of mobile platforms, often a lack of security measures.