Projets novateurs réalisés

Smart Grid Diagnostics using Power Line Communications

Power Line Communication uses the existing infrastructure of electrical power lines for high frequency data transmission. The availability of such power line communication modems across the grid, which are typically used for enabling smart-grid connectivity, also gives us the opportunity to extract insight into the nature of the grid. This includes investigating the health of the power cables and also detecting and locating possible line faults. In this research project, we adapt theoretical formulations of this concept to incorporate practical types of cable degradations and faults typically encountered in the distribution network. We use machine learning techniques to enable the power line modems with distributed intelligent grid sensing abilities to automatically diagnose the grid by identifying fault locations, assessing the severity of cable degradations, and estimating the remaining service age of power cables throughout the grid.

Upper Fraser River bull trout management evaluation

Bull trout in the upper Fraser watershed (UFW) of British Columbia are important top predators and serve as the basis of a recreational fishery. Anglers in the region have asked government to consider changing current fishing regulations for bull trout from catch-and-release to regulations that allow them to take a portion of their catch home. Allowing for this regulatory change would increase the types of fishing opportunities in the area but could harm bull trout populations. This is further complicated as there is little information on the ability to sustainably harvest bull trout, and as the species has conservation listing. TO BE CONT’D

Structural investigation of a N-his6 variant of cytochrome P450cam (CYP 101A1) by protein NMR studies

Cytochromes P450 are enzymes that add oxygen to organic molecules, to convert them to water-soluble products that can be further metabolized. These oxidative enzymes are ubiquitous (found in animals, plants and microorganisms). We are working on cytochrome P450cam, isolated from a wild soil bacterium. This enzyme catalyzes the oxidation of camphor (an organic molecule that can be used by these bacteria as a sole carbon and energy source) to 5-exo-hydroxycamphor. The enzyme needs oxygen (O2) for this. We discovered an unexpected reduction reaction catalyzed by this enzyme, to give a camphor-reduced product, borneol, that occurs when O2 levels are insufficient for the bacteria to fully digest camphor. This reaction is used by the bacteria to signal that O2 levels are too low to metabolize camphor, which becomes toxic to them if they cannot digest it. I discovered that a variant of P450cam with additional residues, his6-P450cam, does not catalyze the reduction of camphor to borneol. Based on previous work, we hypothesize that, to reduce camphor, the P450cam needs to be in a closed state, as opposed to open states. “TO BE CONT’D”

AI solutions to patient-physician engagement

The goal of Engagement Intellect system (at Deloitte) is to use conversations between patients and physicians and convert them into useful information. This is done using recent advancements in artificial intelligence technologies that can automatically extract symptoms, medical history, and other relevant information from the voice recordings. Besides extraction, the project will focus on associating this with the relevant context such as people and their relations, or time and location information. Such technological advancement in the medical arena holds immense potential not only to improve the treatment, but also to make medical advice more accessible.

Prediction of Optimal Business Structure for Tax Efficiency

With the growing relevance of data, PwC has embraced Data & Analytics as part of its culture. The Tax practice, specifically, strives to bring value to market by empowering domain experts with automation and data-driven services. Among their initiatives, the Tax Technology team is investigating methods to evaluate the quality of corporate tax returns (T2). Given the complex nature of these documents, this problem requires exploring distinct issues:

1. A T2 can contain hundreds of sub-forms supported by large professional teams within distinct specializations. Scalable methods are needed to identify finalized tax returns among drafts and the large volume of work product produced by these teams.

2. A mapping between each field in a tax return and its corresponding sub-form does not exist.

3. Tax forms contain slips—copies of a tax question which must be answered for any number of relevant cases—yielding forms of vastly different structures.

With the University of Toronto’s support, PwC hopes to assess the integrity of its data and develop proof-of-concepts to optimize service delivery, along with identifying practices and procedures that will differentiate them in the marketplace.

Machine learning modelling of temporal enterprise data

In a few words, this is a fintech data analysis project. The idea is, given temporal data of financial nature, to build algorithms that predict its evolution over time. For instance, the data could be certain assets prices, or customer buying history. The objective would be to respectively predict this asset price in a close future, or to anticipate what the customer wants to buy next and make relevant recommendations. Such algorithms belong to the family of machine learning algorithms. In the last years, machine learning has been of keen interest to the worldwide scientific community, as a sub-branch of machine learning named deep learning has seen considerable algorithmic progress. This project would leverage recent and efficient deep learning methods.

Voice Cloning Optimization

An artificial intelligence tool that is capable of generating natural-sounding speech can be embedded into many valuable services such as conversational agents for the disabled and conversational assistants. Such tool, when equipped with the capability of mimicking individuals’ vocal characteristics, will improve personalization of these services. In this project the intern will seek to develop a state-of-the-art voice cloning model. The two major research agendas underlying this project are as follow: to develop a data-efficient voice cloning model with high speech synthesis quality; to enrich voice cloning models with the capacity of voice properties manipulation for speech synthesis. On the former agenda the intern will set out to develop a model that requires minimal fine-tuning to pick up a new speaker’s vocal characteristics and use them to generate realistic audio samples. TO BE CONT’D

Validation and Usability Analysis for a Return to Work Software Platform

Benefit payments totaled 2.5 billion dollars for Ontario workplaces in 2015. The most common injury resulting in occupational lost time claims is a strain or sprain. These injuries indicate that despite massive efforts to reduce musculoskeletal injuries in Canadian workplaces, these issues are still a prominent source of disability and have an associated $2.5 billion annual economic burden. A physical demands description (PDD) database allows health care practitioners to determine if an employee can return to work by comparing their residual functional physical capacities and physical job demands. These PDDs lack format standardization, require technical expertise to perform, and are time-consuming. This research will examine the utility of a new, rapid, and easy-to-use video-based PDD tool, examine its validity compared to traditional pen-and-paper methods, and assess if the new video-based PDDs aide occupational health physicians in determining an appropriate return-to-work plan.

Automated Visual Inspection, Sentencing & Dressing

Within the aerospace sector, aftermarket services account for over 50% of revenue generated by aero engine manufacturers. Central to this is the ability to inspect and repair high unit cost components. Many processes are manual but given the ever-increasing quality, cost and delivery requirements, and the safety critical nature of these rotating parts, there is a strong drive towards process automation. The objective of this project is therefore to productionise and validate the automation of inspection, sentencing and removal of defects present on service-run components such as gas turbine discs, shafts, blisks and fan blades. Each of the partners are expected to be advantageous to the project because of their reputations preceeding them. For AV&R, the potential for future system deployment is huge through Rolls-Royce sites and joint ventures.

Caractérisation et valorisation de la fibre du bois et de la biomasse contaminée

Le développement économique de plusieurs régions du Québec et du Canada dépend à la fois d’une utilisation rationnelle du matériau bois et de la valorisation de toute sorte de matières bioressourcées. Le projet de recherche propose d’explorer différentes technologies de caractérisation et de valorisation de la fibre du bois et de la biomasse contaminée. Plus spécifiquement, ce projet vise à développer et évaluer la caractérisation des attributs de la qualité de la fibre en utilisant des outils non destructifs à la fine pointe de la technologie (résistographe, ultrason et analyse vibratoire); de valoriser la biomasse résiduelle contaminée par traitement thermique (technologie CarbonFX d’Airex) dans le cas du bois traité et décontamination électromagnétique des métaux dans le cas de l’industrie de transformation; et de développer des nouvelles alternatives de valorisation de la fibre du bois: TO BE CONT’D

Tokens, cryptocurrencies, and cash: new forms of currency and the establishment of their value in the platform economy

This project explores the use of new forms of currency and revenue models in the context of online platforms, looking for their different economic and symbolic implications. By looking at an adult website that broadcasts amateur sexcams, this research asks how money – tokens, crypto currencies and traditional currencies, are experienced there, querying their different accounting, representational, and relational systems currently in place. This investigation will give an important insight on the new conditions and practices of digital labor and their alternative revenue models, contributing to a better understanding of the economy of digital platforms. Through an emphasis on the new types of monetary exchanges, this research will help as well to the comprehension of the role of new technologies in the extraction of value from the private sphere.

Democratic dance parties as ‘safe-houses’ for inter-cultural sharing in the context of the refugee crisis

My research looks at emergent possibilities for certain democratized shared spaces and the humanizing potential of cross-cultural sharing in the context of the current refugee crisis. I will do this through a case study of a project currently underway in Munich, Germany, called Plug in Beats. Plug in Beats is a monthly dance party held at Feierwerk, a not-for-profit community arts center with a mandate to bring different people, and cultures together in order to shape dialogues at the interfaces between culture, business, politics, and social issues. Plug in Beats’ concept requires some explaining. The principal is that party-goers arrive with their mobile phones, and are given a number. The numbers are pulled at random by the DJ, who will play a selection of music from the cell phone of the person who’s number has been drawn. Each number may only be drawn only once. As such,the selection process mitigates for tendencies of domination or exclusion of any one style, person or group. The party has been extremely successful in bringing together migrants and ‘local’ Munich residents all with diverse geographical histories and identities. “TO BE CONT’D”