Statistics for geometric data

Overview

• Date: December 17-21, 2018
• Venue: FRIAS, Albertstraße 19, 79104 Freiburg im Breisgau
• Conference hotel: The Alex, Rheinstraße 29, 79104 Freiburg im Breisgau
• Organizers: Philipp Harms (Freiburg University) and Matthew Reimherr (Penn State University)
• Funding: We gratefully acknowledge support of the Freiburg-Penn State Collaboration Development Program 2018, which is co-funded by FRIAS and PSU.

Topic

This workshop is part of a collaboration between the anthropology, statistics, and stochastics departments in Freiburg and Penn State on statistics for geometric data and applications to anthropology. With the proliferation of affordable biomedical imaging technologies, the statistical analysis of 3D shapes has become an important task in biological sciences. Ignoring their inherent structures and complexities can bias scientific conclusions, especially when there is large shape variability. This workshop addresses a blend of fundamental theoretical questions and applied problems with the goal of expanding the fields of functional data and shape analysis, focussing on anthropological applications.

This is a follow-up to the previous workshop at Penn State University in June 2018. 

Participants

• Karthik Bharath (Nottingham University)
• Peter Claes (KU Leuven)
• Philipp Harms (Freiburg University)
• Roos Hopman (University of Amsterdam)
• Hyun Bin Kang (Penn State University)
• Bharath Sriperumbudur (Penn State University)
• Elodie Maignant (Université Paris Saclay; via Skype)
• Matthew Reimherr (Penn State University)
• Stefan Schlager (Freiburg University)
• Thorsten Schmidt (Freiburg University)
• Mark Shriver (Penn State University)
• Anuj Srivastava (Florida State University)

Preliminary schedule

Standard location: seminar room on the 2nd floor or coffee lounge on the first floor at FRIAS, Albertstraße 19, 79104 Freiburg im Breisgau.

• Monday
  • 10:00-12:00 Peter Claes: Convolutional neural nets for surfaces
  • 13:00-14:30 Roos Hopman: An inquiry into ‘the face’ generated through forensic DNA phenotyping
  • 15:00-16:00 Philipp Harms: Introduction to Riemannian shape analysis 
  • 16:15-18:00 lecture by nobel laureate T. Südhof on "Deconstructing the Molecular Logic of Neural Circuits: Cell-Adhesion Molecules and Beyond." Anatomy lecture hall at FRIAS, Albertstr. 19.
  • After the lecture: Christmas market
• Tuesday
  • 9:00-10:00 Brief introductions to the research subjects of the participants
  • 10:00-11:00 Matthew Reimherr: Comparison of methods from functional data analysis and shape analysis
  • 11:00-12:00 Peter Claes: 3D facial analysis
  • 12:00: lunch at Schwarzwälderhof, Herrenstr. 43 
  • 14:00-18:00 multidisciplinary discussion on DNA-based facial reconstruction
• Wednesday:
  • 9:00-10:00 Karthik Bharath: A comprehensive setting for elastic shape analysis
  • 10:00-11:00 Matthew Reimherr: Regression methods for manifold-valued data
  • 11:00-12:00 Mark Shriver: Challenges in the analysis of 3D faces 
  • 13:30-14:30 Anuj Srivastava: Drastic consequences of PCA for misaligned surfaces
  • 14:30-15:30 Matthew Reimherr: Differential privacy for surface-valued data
  • 15:30-18:30 Group work
• Thursday:
  • 9:30-12:00 Discussion on kernel methods for manifolds
  • 9:30-12:00 Discussion on the implementation of deformetrica
  • 12:30-14:30 (optional): reading seminar on heat kernel expansions (room 119 in Ernst-Zermelostr. 1)
  • 14:30-17:30 Group work
  • 18:30: conference dinner at Primo Market (Bernhardstraße 6, 79098 Freiburg)
• Friday
  • 9:30-12:00 Discussion on future research directions
  • 13:30-16:00 Discussion on kernel learning in shape analysis and neural nets in shape analysis
  • 16:00 Skype with Elodie Maignant