[Legacy Report] Computer Vision Series Talk: A Geometric Solution for an Innocent-looking Problem in Metric Learning

#Riemann #geometry #Machine #learning #Metric #Dimension #reduction

Speaker: Dr. Mehrtash Harandi

Mehrtash is a senior research scientist (associate professor equivalent) of the computer vision research group in Data61-CSIRO, interested in various aspects of of learning, especially with a flavor of visual data.

Abstract:

To be tractable and robust to data noise, existing metric learning algorithms commonly rely on PCA as a pre-processing step. How can we know, however, that PCA, or any other specific dimensionality reduction technique, is the method of choice for the problem at hand? The answer is simple: We can’t! To address this issue, we introduce a unified formulation for dimensionality reduction and metric learning by making use of the concepts of Riemannian geometry.

• Add Event to Calendar
  iCal
  Google Calendar

• Acton (Australian National University), Australian Capital Territory
• Australia

---

• Contact Event Host
• Co-sponsored by Shaodi You