Detecting Key Concepts from Low-Quality Data for Better Decision (2023-2026)
Abstract
The project aims to develop data analytics techniques that aid better decision making in high-stake scenarios when data are less-trustable. While data-aided decision making has been widely used, less-trustable data may significantly distort the decisions made and hurt people impacted by these decisions. The outcome of this project expects to be a series of techniques covering data understanding and enhancement, model development and fitting, and novelty detection, to reduce the damage of less-trustable data. The research expects to benefit the people and companies impacted by data-aided decision making in cybersecurity, healthcare and financial fraud detection, providing risk-control services.
