Fast and Accurate Computation of Wasserstein Adversarial Examples

Machine learning (ML) has recently achieved impressive success in many applications. As ML starts to penetrate into safety-critical domains, security/robustness concerns on ML systems have received lots of attention lately. Very surprisingly, recent work has shown that current ML models are vulnerable to adversarial attacks, e.g. by perturbing the input slightly ML models can be manipulated to output completely unexpected results. Many attack and defence algorithms have been developed in the field under the convenient but questionable Lp attack model.

Effectiveness of vegetation and habitat characteristics as predictors of insect parasitoid populations - Year two

Climate change, land development, invasive species, and other disturbances can alter the composition, structure, and functions of native vegetation across landscapes. These disturbances also impact insect parasitoids, which are a key, and often overlooked, component of biodiversity. By their ability to control other insect populations, they are integral for fostering resilient and functional forests.

Quantifying the value and risk of restoring wetland habitats in agricultural landscapes - Year two

Wetlands provide critical habitat and valuable ecosystem services. Land use conversion in Ontario, however, has led to substantial wetland loss. The restoration of wetlands on agricultural properties has the potential to offset wetland loss, yet these wetlands are also susceptible to contamination by pesticides.

Development and integration of feature detection algorithms for metal-based direct deposition processes

Metal-based direct energy deposition processes, such as robotic welding and laser powder fed additive manufacturing, ideally require feedback sensing of the deposition quality using vision detectors. Image processing algorithms are challenging to develop due to changing process operating conditions. Despite challenges, implementing in-process image processing algorithms is beneficial for traceability and quality assurance, for calibrating process models, and for developing closed loop control algorithms which are able to maintain deposition quality within acceptable quality margins.

Nanocomposites as Multifunctional Structure for Space Radiation Shielding

During interplanetary travels, space radiations are always crucial to spacecraft design due to their extreme hazards to human beings and electronics. Conventional materials, such as aluminum and other heavy metals, have been widely used on spacecraft. The cost of launching overweighed items to space is still high even reusable rockets are available now. Reducing weight of shielding structures on spacecraft will largely benefit future space missions. Nanocomposites with extreme low density but high radiation shielding properties are proposed in this project.

Printability of H13 tool steel by powder-fed laser additive manufacturing

The scope of this research is to provide a comprehensive recipe for powder-fed laser additive manufacturing (PF-LAM) of H13 tool steel. The H13 components manufactured by PF-LAM can potentially show superior mechanical properties as compared to H13 steel manufactured by conventional methods; however, the process optimization is requisite to achieve the additively manufactured H13 tool steel with satisfactory properties. The process is optimized to acquire the printed parts with minimum defects.

Do existing lending products of credit unions meet the financing needs of co-operatives - The case of Libro Credit Union and Ontario’s Co-operative Sector

Credit unions and co-operatives share a common identity and foundational values as organizations that meet the economic, social, and cultural needs of their member-owners. A perennial concern for the credit union and co-operative sectors has been the need to identify opportunities for collaboration to advance the prominence of member-owned models. This research surveys and interviews co-operatives in Southwestern Ontario to understand their financing needs, and to identify why existing financial products do not meet these needs.

A flexible non-enzymatic lactate sensor with long shelf life to detect anastomotic leaks post-surgery

Development of leaks in abdominal cavity following surgery cause acute complications that have a high mortality rate and lead to expensive corrective procedures. These leaks are mostly detected following the development of complications as currently there is no method for their early detection and hence timely intervention which can save lives is not possible. In this aspect, lactate is a crucial bio-marker that is excreted from the leak site prior to the development of a full breach and hence its monitoring will lead to early detection of such leaks.

Geometric Control Methodology for Modular Residential Construction Systems

As the use of modular construction grows as a favorable affordable housing solution, there is a need to determine which parts of the building are most critical to manufacture with a high degree of precision. Producing a whole building to the same level of precision as an automotive or aerospace assembly for instance, is costly, inefficient and waste of resources (if the goal is to ensure the building functions properly and can be assembled without issues). This project develops an overall strategy for addressing level of precision and optimal geometric control.

Mitigation of product degradation in anesthetic gas capture

This project will assist a company with the commercialization of a system that captures anesthetic gases from hospital air emissions, to reduce their environmental impact and global warming effects. A working system is already developed, but under some conditions the anesthetic gases break down to form a hazardous hydrofluoric acid material. This acid is undesirable and can cause damage to the system equipment over time. This research will identify the root causes and mechanisms for this acid formation.

Pages