Weiqing Gu
Biography
Dr. Weiqing Gu, CEO of Data-to-Decision (Dasion), holds MS and PhD degrees in mathematics and an MS in Computer and Information Science. As a US citizen, she formerly directed the Clinic Program and taught as a Professor at Harvey Mudd College, where she spearheaded over a hundred CS-Math industry projects since 1998. Her research in medicine and the military utilizes ML and big data analytics extensively. In 2007, when the American Mathematical Society selected Dr. Gu and her students to present their NSF-funded research to the US Congress, she showcased the potential of mathematics to aid in curing cancer. Following this, she served as a program director at the NSF, earning an outstanding evaluation for her role in selecting transformative research proposals.
Dr. Gu's career has also encompassed significant contributions to Big Data Analytics for the US Defense Threat Reduction Agency and the US Navy. She is deeply connected within the industry, having collaborated with over 50 companies through Harvey Mudd’s clinic program, including Proofpoint and Virgin Orbit, and has taught over a thousand students at both Harvey Mudd College and Claremont Graduate University. Her summer industry collaborations have included projects with over 30 companies such as UnifyID, Intel, Laserfiche, and Unilever.
Research Interest
Dr. Gu is deeply involved in research for predictive models and anomaly detection in Machine Learning with applications in precision diagnosis. Her former research on the geometry of a manifold (e.g. sphere or Grassmann manifold) and computational geometry applies to fundamental problems in dynamics and control theory
Abstract
Geometric Unified Learning for Neurological Disease Detection
This talk introduces Geometric Unified Learning (GUL), a cutting-edge technology designed to enhance the detection of neurological diseases and brain disorders. By leveraging reusable building blocks and techniques rooted in differential geometry, GUL addresses critical challenges in deep learning, including data overfitting, inefficiencies, and lack of interpretability. It provides clinicians with transparent, trustworthy, and highly interpretable predictions by identifying intrinsic patterns within complex brain structures. GUL's efficient data processing and simultaneous search and learning capabilities enable robust, flexible, and resource-efficient solutions. By reducing the time and effort required for data analysis, GUL not only supports better clinical decision-making but also saves valuable time for healthcare providers, ultimately improving patient outcomes. This approach tackles challenges like data quality, scalability, and parameter tuning, offering a powerful alternative to traditional deep learning models.
