Innovative Projects Realized

Machine learning developer interns within cross-functional teams to develop and commercialize AI-powered solutions – Part 1

AltaML is an innovative company capitalizing on a major technological trend: artificial intelligence (AI) technologies, enabled by big data, are driving a fourth industrial revolution. AI will transform all industries, but traditional industries face challenges in implementing AI. AltaML has a unique business model to overcome barriers to adoption of AI solutions by industry, which is to bring the innovating startup together with the large organization, thereby bringing together rich datasets, AI talent with a playbook for industry application and the close collaboration of subject matter experts and AI experts–with a mindset for change. In addition to AI expertise, we bring agility that our large, corporate partners often lack, and which is so essential for innovation. With a strategic focus on AI adoption and product, AltaML works across industries as well as with the public sector using a co-development approach to create applied AI solutions as well as joint AI ventures. This rich, complex multisectoral environment provides the breadth that enables insights in one area to be applied in new areas, leading to ever increasing opportunities for innovation.

The project comprises internships in a variety of technical and business roles, which are: associate machine learning developer, business development associate, communications associate, finance associate, associate business solutions consultant, and project delivery associate. Outcomes will include algorithm creation and deployment, data visualizations, market research reports, sales collateral, competitive landscape analysis, feasibility analysis; key messages and content writing such as case studies and feature articles, financial model development, data analysis, financial reports and variance analysis, customer workflow mapping, business case reports, resource allocation plans, project update reports, and project plans.

Business interns within cross-functional teams to develop and commercialize AI-powered solutions – Part 1

AltaML is an innovative company capitalizing on a major technological trend: artificial intelligence (AI) technologies, enabled by big data, are driving a fourth industrial revolution. AI will transform all industries, but traditional industries face challenges in implementing AI. AltaML has a unique business model to overcome barriers to adoption of AI solutions by industry, which is to bring the innovating startup together with the large organization, thereby bringing together rich datasets, AI talent with a playbook for industry application and the close collaboration of subject matter experts and AI experts–with a mindset for change. In addition to AI expertise, we bring agility that our large, corporate partners often lack, and which is so essential for innovation. With a strategic focus on AI adoption and product, AltaML works across industries as well as with the public sector using a co-development approach to create applied AI solutions as well as joint AI ventures. This rich, complex multisectoral environment provides the breadth that enables insights in one area to be applied in new areas, leading to ever increasing opportunities for innovation.

The project comprises internships in a variety of technical and business roles, which are: associate machine learning developer, business development associate, communications associate, finance associate, associate business solutions consultant, and project delivery associate. Outcomes will include algorithm creation and deployment, data visualizations, market research reports, sales collateral, competitive landscape analysis, feasibility analysis; key messages and content writing such as case studies and feature articles, financial model development, data analysis, financial reports and variance analysis, customer workflow mapping, business case reports, resource allocation plans, project update reports, and project plans.

Novel fuel cell materials and designs for high performance

Polymer electrolyte membrane fuel cells are a promising solution to addressing climate change, as they can produce emission free energy on demand using hydrogen as fuel. Despite their benefits, many challenges remain in the way of widespread commercialization of these devices. Specifically, components such as the gas diffusion layer (GDL) and catalyst coated membrane (CCM) suffer from poor mass transport and durability issues. These components must be designed for durability and improved transport properties to become commercially viable. To address these challenges, we propose to apply both pore network modelling and advanced electrochemical characterization techniques to reveal the mechanisms responsible for degradation in the performance of the GDL and CCM. The insights gained from this collaborative work with Prof. Auvity will inform the design of next-generation fuel cell porous media for clean energy applications.

Post-discharge health-related quality of life, disability, and risk of post-traumatic stress disorder amongst the survivors of the veno-venous extracorporeal membrane oxygenation during the COVID-19 pandemic

During the COVID-19 pandemic, patients with severe pneumonia sometimes required a type of life-support
machine for their lungs called ECMO. Using large intra-venous lines, ECMO removes blood from a patient, adds
oxygen to the blood, and puts it back into the body. While it can be life-saving, some patients can develop a
brain injury while on ECMO. However, the long-term effects on the brain of being on an ECMO machine are not
known. Some patients may experience difficulties with high-level thinking, or emotional difficulties like
depression or post-traumatic stress disorder. The purpose of this study is to examine cognitive and emotional
function of patients who have survived ECMO. We will contact patients after they have been discharged home
and ask them to complete tests of cognitive and emotional function. We can then use this information to see
which patients may be at risk of developing these issues, so we can better support them in the future.

A framework to enhance deep learning systems’ trustworthiness against Out of Distribution examples

In the past decade, deep learning models have demonstrated their highest performance for a variety of tasks. These models outperformed classical machine learning models and even humans in terms of performance and accuracy. However, previous research indicated that these models are vulnerable to out-of-distribution and adversarial inputs. Ideally, these inputs should be rejected by the deep learning model, but the deep learning model generates confident outcomes for it. In this research, we develop a framework that assesses the deep learning model’s vulnerabilities against such inputs, detects and rejects these malicious input , and enhances deep learning models to generate less confident labels for them.

Efficient Emitters for OLEDs using Structural Constraint

Lighting is one of the largest demands on the global electricity supply at nearly 20%, meaning that a small change to more efficient lighting technology can have an outsized impact on our environmental footprint. Current research into efficient lighting technology is focused on third-generation organic light emitting diodes that emit light through thermally activated delayed fluorescence. Although devices with promising efficiency have been demonstrated, there are still barriers to commercialization of this technology. One of the major barriers is that blue emitters are lagging behind other colours in performance and stability. Blue emitters require higher energy excitation compared to other colours which can cause degradation over time leading to poor device lifetimes. In this project, we present a new molecular design that will allow high performance blue emitters to be targeted by modification of a common phenothiazine donor. Adding bridging groups will improve the stability of the molecule, while donor oxidation can tune the emission colour to target stable display-quality blue materials.

Unions and the Macro-Economy

This research proposes to examine the relationship between unionization in Canada and Canadian rates of employment, unemployment, and the distribution of earned income. Particular attention will be paid to the institutional environment in which unions and collective bargaining exist. The problem to be explored is the way in which the institutional environment (such as laws governing collective bargaining rights) has shaped the effect of unionization on the Canadian macro-economy and consequently affected the employment and income opportunities of working Canadians

Cascades : Explicabilité des anomalies de séries temporelles industrielles

Les ruptures de feuilles de papier produites par Cascades sont une préoccupation majeure pour Casades car elles diminuent la fiabilité et l’efficacité du processus de production de la pâte à papier, entraînent des coûts substantiels et posent un risque important pour la sécurité des opérateurs de l’usine [3]. Par conséquent, ce projet aborde le problème de la production robuste de pâte à papier en utilisant les techniques d’analyse et d’apprentissage automatique.
L’objectif ici est de parvenir à stabiliser le processus de production des machines employées par Cascades en identifiant les valeurs aberrantes des variables pouvant causer une détérioration de la qualité du produit, de la vitesse de production, et des risques de rupture de feuille. Le but est donc d’améliorer le taux de rendement global en proposant des explications plausibles sur les déviations temporaires de production afin d’améliorer les contrôles employés dans le procédé.

Modelling Light Transmittance for the Human Heart using Computational Modelling Techniques in the Effort to Apply Optogenetics for Heart Defibrillation.

Optogenetics is a new and rapidly growing field of bioengineering that allows to control physiological functions of genetically modified cells using light. Applying this novel technique in cardiac applications opens the door to the development of the next generation of cardiac devices, such as contactless implantable cardioverter defibrillators (ICDs). A challenge with the potential use of optogenetic devices in human clinical applications is the limited light penetration through cardiac tissue. In this research, the optical properties of cardiac tissue are experimentally measured for preserved and fresh porcine hearts. The results obtained are used to develop a porcine heart model, which will be further modified and extrapolated to develop a human heart model. The model of light transmittance through human cardiac tissue will serve as a tool to simulate the effectiveness of optogenetic defibrillation. This research is expected to advance the theoretical knowledge in the field of optogenetics and benefit the efforts to develop an optical-based ICD that can be used in clinical settings to improve patient care.

Pathology and disease diagnosis using artificial intelligence and machine learning: Supervised and unsupervised deep-learning based methods on data from medical imaging procedures and patient chart analysis

Machine learning applications in healthcare have shown excellent inroads in medical imaging sciences in recent years. Our research aims to improve upon and open up doors into several different pathology diagnosis applications using artificial intelligence. Contemporary research into some of these applications has shown better diagnostic capacity than expert-level clinicians. Our research into artificial intelligence applications in healthcare include autonomous polyp and bone metastasis pixel-level detection, along with pathology detection in chest x-rays without explicitly labeled data. Through a commercialization research program with Lab2Market, we aim to figure out how exactly our technologies can be improved upon and leveraged to improve clinical outcomes for patients and figure out how we can meet their needs through interacting with our potential customer base.

Biopolymer-integrated pultruded glass fiber-reinforced (FRP) composite materials for shoreline protection

As part of a larger proposal, the proposed research project aims to develop biopolymer-integrated, environmentally-resilient thermoplastic glass fiber-reinforced polymer (FRP) composite materials with the goal to protect Atlantic shoreline from soil erosion due to accelerated climate change events. The project, in partnership with New Brunswick Innovation Foundation (NBIF) and Thermopak Ltd., will modify the conventional pultrusion manufacturing methodology to work with a novel biopolymer (developed by UNB Nanocomposites and Mechanics Laboratory) and manufacture composite rebars in an in-house pultrusion machinery, followed by technology transfer to Thermopak for industry-scale composite sheet pile development for the stated application.

Hortau : Cartographie d’un indice de stress hydrique basé sur des valeurs de potentiel matriciel du sol

Hortau développe des solutions technologiques innovantes pour améliorer le suivi de l’irrigation des cultures
agricoles. L’un des défis les plus importants dans l’optimisation des pratiques agricoles est de bien mesurer en
fonction du temps la quantité d’eau disponible dans le sol afin d’adapter l’irrigation de façon journalière. La tension
du sol est une métrique qui permet d’évaluer l’énergie qu’une plante doit utiliser pour puiser son eau du sol. Hortau
utilise des tensiomètres pour calculer, avec une très grande précision, la tension du sol, mais cela demande
beaucoup de senseurs dans le sol. Ce projet cherche à réduire le nombre de senseurs nécessaires tout en gardant
une certaine précision. En utilisant les données que Hortau possède, nous voulons faire une extrapolation des
valeurs entre les différentes stations. Cette information contribuera au développement d’un modèle
d’apprentissage automatique qui permettra aux producteurs de prendre des décisions plus éclairées pour
améliorer leur rendement agricole en contrôlant les périodes optimales et les niveaux d’irrigation.