Innovative Projects Realized

Panoptic segmentation of an underground mine point cloud

Underground mining operations highly depend on accurate information for intelligent decision-making, may it be safety-wise or efficiency-wise. However, the industry still uses old technologies to keep track of the evolution of their mines. This project aims to develop a new technology to help keep mines safer and make operations more efficient using point clouds. Point clouds are a novel means for 3D visualization in which space is mapped through points using specialized scanning equipment. For this project, various algorithms are developed to harness the power of point cloud visualization. The key goal is to identify structures (e.g. ventilation ducts, piping, etc.) in such 3D visualizations, through the use of artificial intelligence.

Modeling, simulation and optimization of municipal solid waste gasification process through physics informed deep learning

Municipal solid waste (MSW) refers to recyclables and compostable materials, as well as garbage from homes, businesses, institutions, and construction and demolition sites. Disposal of MSW causes significant environmental problems. It is imperative to develop efficient environmental-friendly treatment technologies to tackle this global challenge. Among feasible technologies, gasification of treated MSW has been considered as a critical option since it could carry out the waste prevention and energy recovery from waste. However, the characteristics of MSW are hardly investigated due to complex organic matters, moisture content, carbon, nitrogen, and sulphur. While modeling and simulation of MSW gasification process is critical for controlling and optimizing the process operations, the traditional physics principle based process model is computationally expensive and fast meta-models are required. As the artificial intelligence (AI) technologies became one of the major defining attributes of competitive advantage across many manufacturing processes, this project aims at the development of physics-informed machine learning model for the prediction of syngas composition, gas production rate and heating value of gas produce in MSW gasifiers. The developed model will leverage the power of recent development in artificial intelligence to support the development of advanced MSW gasification process.

Biochemical Studies and Assay Development Targeting Novel Anti-cancer Agents

Cancer is a worldwide health problem and is the leading cause of death in Canada; as such, novel treatment strategies are constantly being sought. Compared to normal cells, cancer cells exhibit aberrant metabolism characterized by increased glucose uptake and rapid growth. The development and biochemical analysis of a series of small-molecule drug candidates that possess the ability to selectively disrupt tumour metabolism will lead to a better understanding of cancer cell metabolism as a whole and ultimately, new treatments for cancer. Development of this oncology program will increase the innovative technical capacity of Alectos Therapeutics and expand its capabilities for future drug development programs.

Decryption Failures in a Quantum World

The development of scalable quantum computers threatens the secrecy of communication by breaking classical encryption schemes. In order to ensure long-term security in communication networks quantum-safe cryptography is currently being developed and evaluated in standardization competitions.

A special property of many of the proposed cryptosystems is a low but non-zero probability of failure: An encrypted message may fail to decrypt successfully. This decryption depends on a secret key, which is to remain hidden to ensure secrecy of the communication. However, the mere event of such a failure has been shown to leak information about the secret key.

We analyze the vulnerability of quantum-safe cryptosystems to attacks based on these failures.
In particular, we construct a novel model of an adversary’s chance to trigger and exploit such failures on a quantum device.
As a result, new insights into protecting against quantum-attacks that enable the development of future quantum-safe encryption are collected.

Improving Resuscitation Equity using AI

For urgent, life-threatening medical conditions such as cardiac arrest and stroke, treatment is time-sensitive and must be received promptly to maximize the likelihood of patient survival. Previous research has shown that areas with lower socioeconomic status have higher rates of cardiac arrest incidence as well as lower rates of receiving timely treatment and patient survival. This suggests that current practices have an inherent inequity of care across socioeconomic levels.
Artificial intelligence methods present new opportunities for emergency medical service (EMS) systems to optimize their response by identifying cardiac arrest patients sooner and strategically deploying EMS resources to lower response delays. However, the effects of these methods with respect to the equity of care across socioeconomic status is unclear.
The proposed project will be conducted with the Scottish Ambulance Service to develop equitable, AI-driven resource allocation policies such as the placement of public defibrillators and recruitment of community-based responders in ways that can optimize both the effectiveness and the equity of care for cardiac arrest patients, and compare the impact that these policies can have to that of existing practices.

Comparative genomic analysis of Bradyrhizobium strains with natural variability: efficiency and competitiveness implications

A major challenge facing farmers is obtaining the nitrogen needed to support plant growth. Inoculation of legume crops with rhizobia, which supply plants with the required nitrogen, is a green alternative to environmentally hazardous nitrogen-fertilizers. However, rhizobia vary in their effectiveness, and the success of inoculants depends on their ability to outcompete native and poorly-effective rhizobia. Thus, development of highly-competitive and highly-efficient rhizobium inoculants is a key challenge to ensuring rhizobium inoculation becomes a cornerstone of sustainable intensification of agriculture. Here, we will use genomic approaches to characterize the genomes of 12 natural variants of the soybean symbiont, Bradyrhizobium japonicum. Despite these 12 organisms being closely related, their symbiotic properties vary. We will assemble and compare the genomes of the 12 strains, looking for genetic variations such as gene gain, gene loss, and nucleotide substitutions. Results will allow us to identify the genetic basis for the phenotypic differences, laying the groundwork for future work engineering elite Bradyrhizobium inoculants.

Detargeting Protein-Protein Interactions For Cellular Design Applications, Using 3D Structure-Based Deep Learning Models

Rational protein design has had a tremendous impact on pharmaceutical, agriculture, and chemical industries over the past 30 years, by focusing exclusively on individual proteins and their intrinsic activities. The next generation of protein design tasks will seek to modify function inside living cells, competing and interacting directly with pre-existing cellular machinery. Modifying systems in living cells will open a new wave of biotechnology applications, such as living drug implants and diagnostic tools. However, effectively introducing new or engineered proteins into a living cell system requires attentive coordination to ensure that designed protein surfaces do not inadvertently interact with other host cell proteins and disrupt otherwise vital activities. Biological pathways are also very sensitive to perturbation, which can often result in cell death or functional failure. This partnership aims to build a predictive computational technology to ensure that designed protein surfaces do not display recognition features of proteins in the host cell that will minimize these unwanted interactions.

Accelerating quantum variational methods using Bayesian optimization

Quantum computers hold the capability to solve problems that are out of the computational reach of classical computers. The race to finding and testing such problems on a smaller scale is currently on. Variational methods are one of the candidates leading this race. These methods are hybrid, meaning that they require quantum as well as classical computers working together. As the size of the problems increases and becomes relevant to the current needs of society, they also become more challenging to solve. This project is about improving the variational methods by combining them with a special technique known as Bayesian optimisation. The combination will accelerate variational optimisation by reducing the required quantum computational resources. Ultimately, this project will contribute towards solving the problems of interest to society, e.g. simulating large molecules, faster and more reliably once a powerful enough quantum computer is operational.

Soil microbial community in cropping systems integrated with Urochloa spp

Nitrogen (N) is essential in agriculture, but has caused environmental pollution problems, demanding improvements in N use efficiency by crops. Tropical grasses like Urochloa spp. conserve N by biological nitrification inhibition (BNI) and mitigate N losses in soils. However, little is known regarding the effects of BNI plants on soil microbial diversity and functions, especially those related to nitrification. This project aims to assess the effect of cultivation of Urochloa spp. on the soil microbial community in grain crops (maize and soybean). The soil microbial community’s structure and predicted functionality will be accessed using advanced molecular tools and biostatistics. We expect to understand the effects of BNI plant Urochloa spp. on the soil microbial diversity and functions, especially those related to N transformations in soil. These results will support further studies on soil management practices for achieving more sustainable production systems in terms of N dynamics.

Optimisation des flux logistiques dans une entreprise agroalimentaire

L’entreprise québécoise YourBarFactory, spécialisée dans la conception et la fabrication de barres de céréales sans allergènes, connait ces dernières années une forte croissance. Cette croissance s’est matérialisée fin 2018 par la construction d’une nouvelle usine. Afin d’anticiper la pleine utilisation future de ce nouvel outil, il est temps d’étudier l’efficacité des processus, et notamment la manière dont sont gérer les flux logistiques de la compagnie.

Dans un premier temps, nous procéderons à l’analyse des processus logistiques afin de déterminer les points bloquants. Par la suite, nous travaillerons à la rédaction d’un plan d’action proposant des solutions concrètes à ces problèmes. Pour finir, nous mettrons en place ces actions de manière concrète sur le terrain et évaluerons l’impact de ces nouveautés sur les flux logistiques de l’entreprise.

Addiction Neuromodulation System

In the prior Mitacs funding cluster, a clinical trial of opioid addiction was initiated and approved by both Health Canada and the ethics board of the Saint Joseph General Hospital.
This research project expands upon that work. A smoking cravings reduction program using SmartStim transcranial direct current stimulation (tDCS) has been initiated by NorDocs, and Quit4 (an established smoking cessation program) is an early adopter. The projects in this cluster all focus upon optimizing and studying this new business. Each intern’s project is related to investigating this novel paradigm for reducing cravings with tDCS, and investigating the details of the efficacy of this program. The benefits of this project to NorDocs include 1) the potential for improved adoption of SmartStim technology by customers, 2) a better understanding of how to improve procedures related to SmartStim

The Future of Teamwork: Leveraging AI-Based Chatbots to Enhance Organizational Collaboration

Teamwork is at the heart of most organizations today. Many organizations (four in five) have begun to use Enterprise Social Media (ESM) platforms in an attempt to increase the effectiveness of communication and collaboration between their employees. Yet, many challenges remain, such as the lack of engagement with ESM as well as the lack of adequate support of ESM for collaborative work. This research project aims to map out the challenges associated with team communication and collaboration in the context of ESM and leverage Artificial Intelligence (AI)-based chatbots to enhance such team communication and collaboration in organizations. Based on the analysis of an extensive ESM dataset, we identify opportunities for chatbots to support team communication and collaboration activities. Subsequently, we design, develop, and implement chatbots for these activities and integrate them into an ESM-based system. Finally, we analyze how these chatbots affect team communication and collaboration in ESM and, in turn, various individual, team-level, and ultimately organizational-level outcomes. Our research project will generate novel findings on how companies can leverage AI technology in the context of ESM platforms and provide design guidelines for chatbots that support team communication and collaboration.