Innovative Projects Realized

Intelligent analytics for hyperspectral image dimension reduction

This research project focuses on improving the analysis of hyperspectral images, which capture images with many narrow spectral bands. Hyperspectral images have many challenges due to their high dimensionality and data redundancy. To overcome these challenges, the researchers aim to develop an advanced intelligent analytics framework for hyperspectral dimension reduction (HDR) using disentangled representation techniques. This approach will extract low-dimensional features that contain the most important information in a compact and interpretable manner. The researchers will investigate and design a new HDR framework that considers spatial-spectral heterogeneity in HSI using advanced deep neural network architectures and spectral models. They will also adapt and extend the HDR methods to improve various HSI processing tasks such as denoising, visualization, and classification. The expected outcome is a practical software system for enhanced HSI interpretation in real-world applications.

Algorithmic redaction of refugee court files

The goal of this research The system will be designed to intelligently propose redactions for court officers to examine, thereby integrating human oversight with advanced technological capabilities.is to engineer an intuitive and user-friendly system capable of automatically scrutinizing and eradicating confidential information contained within refugee court files in Canada. By leveraging cutting-edge language processing and machine learning methodologies, the project seeks to revolutionize the way legal documents are handled. This innovative approach is centered on preserving the privacy and enhancing the safety of the involved individuals, a critical aspect often overlooked in the rush of legal proceedings. This synergy will facilitate better decision-making, ensuring that only the most pertinent and non-sensitive information remains accessible.Ultimately, this project has the potential to make a profound difference in the lives of vulnerable refugees, offering them an added layer of protection. Concurrently, it upholds the principle of transparency within the legal system, striking a balance between public access to court files and individual privacy rights. This delicate balance is essential in a democratic society and the successful implementation of this system could serve as a model for other jurisdictions.

A human-centered approach to the design and implementation of advanced health technologies using Cognitive Work Analysis (CWA)

While there have been several literature reviews on the performance of digital sepsis prediction technologies and clinical decision-support algorithms for adults, there remains a knowledge gap in examining the development of automated technologies for sepsis
prediction in children. Pediatric sepsis is a major cause of mortality of children worldwide. However, there is still a lack of easy-to-use predictive tools that can accurately diagnose sepsis in children. This research aimed to develop an optimal algorithm for supporting early sepsis prediction in children. The results may inform research on identifying relevant predictive indicators best suited for the design of digital technologies in specific use contexts and environments, improvements towards model development for sepsis prediction and factors supporting the optimal workflow integration of digital prediction systems by clinicians.

Building and testing a simulator of human-computer interface for small modular nuclear reactors

The project is dedicated to the explore how the effective interface for a small modular nuclear reactor should look considering human factors and current digital technology design trends. The research will review the significance and application of modern human-computer interaction techniques to ensure a better user experience and to achieve substantial safety and operational benefits in small nuclear power station workplaces. Based on the findings the report will be prepared summarizing the literature review and design requirements for the SMR interface. The goal of the project is to design and develop a prototype interface for small modular nuclear reactors using appropriate tools and technologies following human-computer interaction principles.

Individual problem solving: experiments on open-ended problems

Open-ended problem solving plays an important role in industrial innovation, product development, strategy formulation, and many other applications. The study will use experimental methods to investigate the behaviour of individuals while solving open-ended problems, and will entail the gathering and analysis of both qualitative and quantitative data. Individual problem-solving tasks will be given to participants with varied degrees of competence, and the outcomes can then be compared to results from previous studies of group problem-solving. Data will be gathered via observation, surveys, and interviews, and will be subjected to statistical and thematic analysis. This study’s expected results include understanding of how different degrees of open-endedness affect individual problem-solving behavaiour and performance. These insights can guide future methods and strategies to solve a variety of open-ended challenges and can help choose the best method for doing so.

Data-Driven and Synthesis-Guided Generalization in CHC-Solving

The intern will be working on a research project aimed improving scalability and applicability of automated reasoning tool called SPACER by improving the way it deals with special data types. Overall, the goal of this project is to make SPACER a more effective and reliable tool for verifying program correctness.

Mitigation of Fouling of Tertiary UF Membranes at Low Temperatures

Membranes that are used in wastewater treatment have been found to clog more rapidly at cold temperatures. This study will examine alternative operating strategies that will reduce clogging and thereby reduce the needs for extra energy and chemical consumption under these operating conditions.

Evaluating the effects of motion cues in virtual truck driver training

The use of virtual reality (VR) and realistic real-time graphics has become a critical component in various industries, including aviation. However, the benefits and necessity of full-motion simulators in training remain unclear, especially when considering the fundamental differences between tasks such as flying and driving. This research aims to address the knowledge gap in understanding the effects of motion cues in virtual truck driver training. While studies have been conducted on motion cueing and its effects on training in simulators for different tasks, there is a scarcity of research exploring the impact of motion stimuli on VR-based truck driver training.

Assessing the effect of mowing practices on stem-dwelling arthropod assemblages in an urban conservation project

This project aims to evaluate the effect of mowing on the insect community living in meadow plant stems. Mowing is a necessary practice in meadow restoration in cities due to the need to control invasive plant species. However, past studies have shown that mowing can also contribute to increased mortality rates in a wide variety of animal groups. This presents a potential trade-off between goals for managing the plants in meadows and goals for managing animals. This project will examine the effect of mowing on insects. Many insects rely on their host plant’s structural integrity to complete their life cycles and may thus be negatively impacted by mowing. A better understanding of the potential trade-off between plant and animal conservation management will allow practitioners to take into account a larger breadth of living organisms and to better inform future mowing practices in urban meadow restoration projects.

Vortex Identification Using Machine Learning

Many fluid flows are dominated by the dynamics of vortex formation and convection. Examples of practical importance include flows over aircraft wings/wind turbine blades and environmental flows. Knowledge of vortex parameters such as position, radius, circulation, and convective velocity are needed to understand and predict the influence of vortices on flow development. Although a vortex is intuitively understood as a region of fluid with a coherent rotational motion, there is no universally accepted method of defining and identifying vortices in a fluid flow. Furthermore, reliable vortex identification in turbulent flows is made difficult by the presence of random velocity and pressure field fluctuations. Previous work on vortex identification has attempted to overcome these challenges through the application of machine learning techniques to identify and track vortices in experimental data and numerical simulations. The goal of this project is to develop a robust method for vortex identification, quantification, and tracking based on machine learning techniques.

Detecting Phishing Websites using Machine Learning Techniques

The project “Detecting Phishing Websites using Machine Learning Techniques” aims to develop a method that can accurately identify and block malicious websites. Machine Learning algorithms will be used to analyze various website features, such as URL, page content, rank and other indicators to determine if it is a phishing site or not. By identifying and blocking these websites before they can cause harm, the system will save time and resources while also preventing the loss of sensitive information. With the increasing threat of cyber attacks, this project is essential for ensuring enhanced safety of browsing. Leveraging the power of Machine Learning, the research has the potential to make a significant impact in the fight against phishing.

The expected outcome of this project is a report describing a method that can later be used to develop reliable and efficient tool to help individuals, organizations and state agencies protect themselves from falling victim to phishing attacks.

Analysis of the error propagation dynamics of PIV-pressure

This study aims to analyze the dynamics of error propagation in the PIV-pressure analysis and suggests two approaches: I) a formal analysis and II) corresponding algorithms for optimal sensor placement that minimize the error propagation from the measured data to the computed pressure field. By analyzing the Poisson problem together with boundary conditions, the optimal sensor placement can be determined in advance, which would minimize the error in measurements. However, since there may be multiple locations that could be considered as good options, this task can be treated as an optimization problem. Therefore, a machine learning algorithm will be utilized to narrow down the possible cases and identify the best location for the sensor placement. Based on the formal analysis from I), we design algorithms to determine the optimal sensor placement that works for engineering and applications purposes. As a final goal, we want to estimate the optimal number and optimal location of sensors to fully describe the pressure field.