Dr Duygu Sap

Education

Ph. D. and M.Sc. in (General) Mathematics, Faculty of Science & Letters, University of Pittsburgh, USA.
M.Sc. in Electronics and Communication Engineering, Faculty of Electrical & Electronics Engineering, Istanbul Technical University, Turkey.
B.Sc. in Mathematical Engineering, Faculty of Science & Letters, Istanbul Technical University, Turkey.

Brief Academic History

Prior to joining CAMaCS, I was a postdoc at the University of Oxford (first, at the Department of Engineering Science; then, at the Mathematical Institute). During my time at Engineering, I was a member of the Impact Engineering Lab and the Solid Mechanics Group, and worked on the ASiMoV project. During my time at Maths, I was a member of the Numerical Analysis Group and funded by the project PRISM. Before I moved to Oxford, I also conducted postdoctoral research in Berkeley as a member of the Research Initiatives Group at the International Computer Science Institute.

Please visit my personal website for more about me and my research.

Research Areas

Geometric Deep Learning, Graph Neural Networks, Graph Attention Networks
Numerical Analysis, Finite Element Exterior Calculus, Discrete Differential Geometry, Applied Algebraic Topology, Topological Data Analysis
Fluid Mechanics, Solid Mechanics, Electromagnetism
Interoperability of CAD systems

CAMaCS Projects

3D Visualization via Morgan Panel Maps (MPMs)

This project aimed at supporting strategic decision-making across complex supply chains. I offered strategies and methods to enhance 3D visualization panels and storylines to support this objective via three distinct applications..

Sensor Visualization

This project aimed at applying topological data analysis and statistical techniques to multi-sensor data obtained from a railway system. I derived various correlations of the sensors measuring two distinct quantities and offered physical interpretations of topological signatures of the dynamical system to enable assessing the structural integrity of the system.

Image Categorization and Retrieval

This project aimed at categorizing second-hand garments and retrieving a given garment from large datasets. I introduced a novel approach based on representative models, which capture the key features of each image, and applied it to image sets using GATs. We also utilized this approach for the categorization of second-hand garments with visual transformers.