In AI safety, compliance ensures that a model adheres to operational specifications at runtime to avoid adverse events for the end user. This proposal looks at obtaining model compliance in two ways: (i) applying corrective measures to a non-compliant Machine Learning (ML) model and (ii) ensuring compliance throughout the model’s training process. We aim to achieve the first via removal of gradient information related to features involved in biasing the model.
Humans possess the ability to see objects as having the same color even when viewed under different illuminations. Cameras inherently lack this capability. A process called auto white balance (AWB) has to be applied by the camera to mimic this behavior of the human visual system. AWB is one of the first steps in a series of operations performed on-board the camera as the raw image recorded by the sensor is processed. It plays a crucial role in ensuring that the colors in the final image that is output to the user are correctly represented.
Increasing demand for more reliable electric power requires advanced monitoring systems that prevent equipment failure and outages. The existing technologies used for monitoring the voltage and the electric field in the vicinity of the high voltage devices are bulky and expensive. On the other hand, maintenance of the monitoring devices requires specific safety precautions. In this research project, a small and inexpensive electric/magnetic field sensor is proposed. They are passive and require no source of power.
Cannabis legalization creates a pressing need to improve existing screening methods. Currently, the two devices approved by the office other the attorney general of Canada (i.e. the Drager 5000 and SoToxa) have not been embraced by the vast majority of police forces who deemed both options unaffordable, difficult to use and inaccurate. The present project aims to create a next generation cannabis detection device capable of accurately assessing the blood concentration of THC.
In the era of digitization, the success of an SME significantly depends on the active engagement with other actors (e.g., brand consumers, suppliers, influencers) in their business ecosystem. In this research project, we propose to develop an engagement model based on the business ecosystem network. These models will predict the customer engagement community association and engagement score. Our model will help the partner organization to understand the critical factor of the life cycle for any particular business on their platform.
The realization of the upcoming 5th Generation of telecommunication standards (5G) requires ultra-broadband transceiver frontends that satisfy both power efficiency and linearity requirements. Since power amplifiers (PAs) are the most power-consuming blocks in these frontends, they are optimized to achieve higher efficiency, and hence lower operating cost and carbon footprint. However, high efficiency PAs suffer from nonlinear behaviours that distort the signal quality and reduce their overall linearity.
The current generation of cellular networks, i.e., 4G, made possible multimedia applications such as music and video streaming in the palm of your hand. The upcoming generation of cellular networks, i.e., 5G, enables new types of applications beyond what 4G offered such as augmented virtual reality, Internet of Things, cloud computing, autonomous vehicles, connected health equipment and connected industrial robots. Communications with drones are expected to be one of the important applications of 5G networks.
In this project, a novel Wireless Power Transmission Integrated Circuit (“WPTIC”) system capable of simultaneous power transmission to multiple devices at varying power needs and distances, through objects of various materials and densities are going to be developed. This system can be obtained by an innovative circuits and antennas structure to model and control the near-field electric and magnetic fields, and also far-field electromagnetic propagation. There are two areas in power transmission, far-field and near-field.
Scrawlr is a platform for unconstrained, global interaction with all internet content and users. Scrawlr allows for user evaluation and unconstrained classification of any Scrawlr-hosted or non-Scrawlr content.
Geothermal energy is a promising source of renewable energy and is gradually gaining attention in application in building heating and cooling system. Standing column wells (SCW) are an efficient way of harnessing geothermal energy for such building applications. However, currently rule-based controllers are used for these geothermal heating-cooling systems with simplifying assumptions to avoid the inherent complexities of the system dynamics.