Grants and Contributions

We wonder how symbolic and connectionist AI could converge in the next-generation deep learning algorithms. We approach this question by investigating and developing deep generative neural networks with capsules (multivariate stochastic neurons), which will represent elements of knowledge in capsules and dynamically piece together to produce novelties that make sense.

The proposed project aims at developing lifetime predictive models. A novel transfer learning-based modeling technology will be investigated. With the proposed modeling technology, the models can be built from "big" historic data and adapted to new operational environments, in turn, which makes the predictive models reusable, evolvable, reliable, and comprehensible.

The effects of aluminum feedstock powder characteristics on cold spray (CS) coating adhesion on steel surfaces will be identified for the first time. It will lead to improvement in the fundamental understanding of the buildup mechanisms at the origin of CS deposition in order to optimise the CS coating adhesion properties.

"Better stress management will reduce incidences of broken rail (#1 cause of rail accidents in Canada). Fibre optic sensing is a game changing solution to study rail stress behavior in-situ. This project proposes to deploy this technology at a rail site to test its suitability and gather critical data".

Blade-Vortex Interaction (BVI) is a well-studied phenomenon in Vertical Axis Wind Turbines (VAWTs) which can cause significant reduction in power generation capacity of the turbine. This proposal suggests a solution that has the potential to mitigate this adverse effect and increase the performance of the turbine significantly.

Flutter is a dangerous unstable vibration of flexible structures/aircraft subjected to aerodynamic forces at high speeds. The project will quantify the effect of shock-wave location variations on the transonic flutter speed, by developing an uncertainty quantification and sensitivity analysis framework. It will significantly minimize the number of simulation runs thus reduce the computational and flight testing costs.

Films of highly ordered P3HT crystallites templated on boron nitride nanotube scaffolds will be fabricated and crystallinity and degree of alignment optimized. Thin film transistors and active layers for bulk heterojunction solar cells will be fabricated from the P3HT/BNNT hybrid formulation and their performance tested.

Development of detectors and measurement methods to characterize monoenergetic x-rays produced by synchrotrons, with the aim of providing the basic metrology necessary for clinical diagnostic and therapeutic applications for such beams.

Key limitations of iPSC-derived cells are the persistence of immature phenotypes and variability between iPSC lines. To address these problems, a Synthetic Biology platform that leverages yeast prototyping will be developed to enable accurate post-differentiation manipulation and correction of genome-scale expression profiles to better recapitulate desired endogenous cellular phenotypes.

The purpose of the proposed project is to investigate feasibility of developing a tool which integrates numerical modeling and remote sensing approaches for detecting, monitoring and modeling fate of plastic particles and in particular microplastic particles (less than 5mm in diameter), in the aquatic environment.