Grant: Idea to Innovation

The impact of Machine Learning (ML) is crucial for high stake human-oriented applications, such as medical decisions, healthcare, and investment decisions, that can have compound effects on making decisions. It is essential for the model to be trustworthy, interpretable, and verifiable. To address this need, Pattern Discovery and Disentanglement (PDD) is a generic interpretable machine learning technology developed at the University of Waterloo (UW) that has been successfully applied to practical applications (such as bioinformatics, healthcare, and clinical practices) for the structured dataset. To support Machine Learning (ML) applications in meeting such demand of the next generation of AI, the objectives of this proposal are 1) extending PDD to input more data types, such as mixed-mode multivariate time series for a wider range of applications; 2) working with relevant industry leaders, who provided attached support letters, for demonstrating PDD technology is ideal to solve their critical business problems; and 3) creating a dashboard to visualize interpretable PDD results to deepen business insights. To achieve the objectives and validate the significance of its commercial value, PDD, ICM, and WeMine technologies, we propose to complete the tasks listed in the proposal in 1 year and assess its success by the quantitative measures we set for each of our short-term and long-term goals. Meanwhile, we will work closely with support companies, especially Anooka Health, to avoid the risk of disclosure of the privacy of patients' records.