Inferring Subjective Ratings for In-car Speech Using Objective Measures

This project explores how computer algorithms will be used to predict the intelligibility and quality of in-car speech processed by hearing aids. Hearing impaired listeners graded in-car speech for a set of conditions. The conditions include seating position of talker, seating position of speaker, levels of background noise, and hearing aid processing methods. Each hearing impaired participant graded processed speech using multiple criteria. For each assessment criteria, a function is generated that maps the assessed criteria to the result of each computer algorithm.

Investigating the transmission of C. difficile in Hamilton-area hospitals and designing diagnostic assays for C. difficile

Clostridioides difficile (C. difficile) infection (CDI) is a common disease associated with hospital stays and is caused by a bacteria known as C. difficile. This project will examine transmission patterns of the bacteria and attempt to develop methods for detection of the pathogen in order to reduce incidence of the disease. This will ease the burden of patients on the hospital system, which I will be partnered with.

Efficient Adjoint Sensitivity Analysis of Emerging Nanophotonic Devices

The design optimization process of photonic and optoelectronic devices can be time consuming. The modeling of these electrically-long devices is time-intensive. Accurate time-domain and frequency-domain simulation tools such as the finite difference time domain (FDTD) method and the finite element method (FEM) are usually utilized to model these devices. These methods require fine meshing to achieve accurate results. The fine meshing results in slow simulation time.
Designing photonic and optoelectronic devices requires large number of accurate simulations.

Profiling the therapeutic potential of fungi from Kapoose Creek

Kapoose Creek Wellness (KCW) is building a vertically integrated business to deliver products and services that take advantage of a growing market for the functional and medicinal potential of mushrooms. A key asset in this vision is 320 acres of land within a rainforest on the North Pacific Coast of Vancouver Island where exotic species of fungi abound. Further, KCW has invested in infrastructure, for example, roads, vehicles, housing and, most notably, a substantial research laboratory, in this remarkable setting.

Rapid detection of Septic Pathogens and their AMR Profiles using a reusable microarray based technology platform.

Sepsis is a life-threatening infection of the blood caused by a wide range of bacterial and viral pathogens. Unfortunately most tests today still take days to detect and identify the pathogens and their underlying antimicrobial profiles. Working together with a leading biotechnology company, founded in Toronto, we plan to modify our recently designed algorithms to capture very low levels of circulating pathogens in the blood in a few hours, potentially saving thousands of lives.

Advanced Quantitative Behavioral Models for Asset-Liability, Interest Rate Risk, and Liquidity Management in Deposit-Taking Financial Institutions

Cashflow uncertainty due to customer behaviors poses special challenges to a bank’s ability to accurately forecast its future cashflows, and therefore makes its funding and risk management difficult. In the proposed research, we plan to use cutting-edge machine learning techniques to study the behaviors of bank depositors and borrowers in Canada using an extensive proprietary data sample of the Partner Organization (i.e., EQ Bank).

Deep Learning on Graphs for Natural Language Understanding in Conversational AI

Deep learning has been the dominant approach to Natural Language Processing (NLP) tasks. NLP applications have successfully exploited the usage of Graph Neural Networks (GNNs). Text inputs are usually represented as a sequence of words, meaning most deep learning models utilize Recurrent Neural Networks (RNNs) and Convolutional Neural Networks (CNNs). However, representing text as just a sequence may lose the structural information of the entire text, and input data can keep this structural information as graphs.

Micron scale ultra-low-cost light emitting diodes made using silicon carbide semiconductor material

A next generation of silicon-carbide (SiC) -based high performance microLEDs has the potential to replace existing gallium-indium nitride-based LED used in LED illuminated displays. The development of these SiC microLEDs is the goal of this research in collaboration with display company AVT Solutions Ltd. Silicon and carbide are low-cost earth-abundant minerals with ?1% of the cost of gallium nitride.

Rapid diagnostics for viral and bacterial pathogens

This project will build on our patent-pending rapid COVID-19 test that uses functional nucleic acid and electrochemical biosensor technologies. The project will work on solving the technological and scientific challenges for scaling-up and commercialization of this type of platform, as well as using the COVID-19 test as a base for building a universal pathogen detection platform. The universal platform will make it possible to rapidly develop new tests for different infectious diseases by engineering new functional nucleic acid probes.

Developing a real-time bacteria detection platform using LAMP assisted bioluminescence assay

Water contamination poses a serious risk to health and economic development in Canada and around the world. The ability to inexpensively and quickly detect the pathogen contamination in water will mitigate the risks associated with water contamination. The LAMP-based biosensor described in this proposal will be able to provide these capabilities, and enable fast and accurate pathogen detection in water in all settings. Ecoli Sense Ltd, a pathogen sensor development company, is looking to develop this capability as part of its pathogen detection device lineup.