Security Risk and Control Modeling for Deep Learning using the SAGETEA Methodology

SageTea Software will contribute expertise in working with Smalltalk and the SAGETEA model. This includes demonstrating the current database model and how it works. SageTea Software will also demonstrate its current implementation of Deep Learning libraries on the Python side including Tensorflow, Kibana and Elastic Search. We will provide expertise in the SAGETEA methodology and assist the researcher with developing additional mathematical analysis, software analysis design and coding. We will assist with testing and also supply infrastructure including cloud environments and software tools.

Exploratory Analysis of Alzheimer’s Disease Detection Using Eye Scans

Optina Diagnostics works on detection of Alzheimer’s Disease using hyperspectral eye imaging techniques. This project was an exploratory analysis of data gathered by Optina so far in order to optimize and fit models to most effectively predict cases. Recommendations from the analysis were then handed over to Optina for future work.

Development of an Al first molecular database to accelerate drug discovery

Using simplified language understandable to a layperson; provide a general, one-paragraph description of the proposed research project to be undertaken by the intern(s) as well as the expected benefit to the partner organization. {100 - 150 words)
The project aims to develop a molecular compounds database to accelerate drug discovery. Compounds shared by chemical providers are currently stored in large library files. Due to their size and number, these files are a bottleneck in virtual screening.

Quo Vadis? Ontologies for Geospatial Question Answering and Consumer Behaviour

A geospatial query is a question where the concept of location is necessary for formulating the answer. Furthermore, we are not simply interested in spatial relationships, but also with the ways in which people can possibly move through space given the goals that they want to achieve. We therefore want to predict the behaviour of people moving through urban environments based on observations about their purchases. In this project, we will explore how can models of commonsense knowledge can be used for automated reasoning to answer geospatial queries and to infer consumer behaviour.

Augmenting Movies with Interactive Narrative Agent

Netflix recently released an interactive movie that brings the concept of interactive media from video games into movie form. In contrast to the passive viewing experience with conventional movies, the interaction brings viewers into the movie scenes for an immersive experience. However, existing interactive movies are still scripted. Though there have been proposals towards freeform conversational for games and agent-directed interaction, viewers are still not in control of the story development.

Decentralized Services for Sharing and Searching User Generated Data

The existing model for applications and services on the internet is a centralized client server model where user
information is under the control of the service provider. As such as centralized model, albeit cost-efficient and
easy to maintain, has created dire consequences for humanity. Hence, for the proposed research project, we aim
to provide tools and algorithms for a decentralized and location-aware experience on the Internet.

Parking Utilization Assessment Using Deep Learning

Analyzing parking behavior and usage in large open-concept retail centers enables owners and managers to better understand how their parking facility is being used. Most large, open-concept shopping centers are experiencing a parking oversupply problem. Current parking allocation is inefficient and contributes to urban sprawl, large concrete pads that trap solar heat and a waste of valuable real estate resources. Parking studies are generally conducted on foot using a combination of manual tallying or with ground level cameras used to collect imagery of ingress / egress traffic.

Applying machine learning techniques for demand forecasting in retail

An important component to every growing retail business is demand forecasting which can affect the strategic plans of a business. The impact extends across the business’ function including budgeting, financial planning, price optimization, sales and marketing plans, capacity planning, staff management, risk assessment and mitigation plans.
In this project, we want to apply machine learning technologies to improve the accuracy and granularity of retail demand forecast.

Development of a breeding database for the application of omics technologies in canola variety development

Next generation plant breeding requires bringing together many different types and sources of information about different plant varieties, including their genetic information together with how well they grow in different environments. The best ways to organize and integrate these sources of information within a plant breeding program is not well known. In this project, we will work with Nutrien Ag Solutions to devise and evaluate a breeding database and associated software tools to assist and accelerate their double haploid breeding program in canola.

Optimizing Wall Formwork Shuttering Design Using Prefabricated Panels in Concrete Construction

Determining an optimal formwork shuttering solution is not a well answered problem, relying heavily on the designer’s preferences and past experiences rather than efficient algorithms and standardized procedures. This results in inefficient solutions that increase overall costs and construction time. Globally, formwork is nearly a $6 billion USD/ year industry; even minor improvements in efficiency of design will result in huge cost savings.

This research aims to develop a standardized methodology and identify general heuristics for solving typical wall formwork problems.