Neuromorphic computing, based on a geometric model of the early stages of vision

2- Day Workshop:

Neuromorphic computing, based on a geometric model of the early stages of vision 

Dates: Friday 9 and Saturday 10 October 2015

Location: CASIA, Beijing, China

Speaker:

Prof. Bart ter Haar Romeny, PhD Northeastern University, Shenyang Eindhoven University, Netherlands

Brain-inspired computing is hot. Modern imaging techniques reveal much of funcional mechaniams, and mathematical modeling is on par. This workshop gives a state-of-the-art insight in neuromorphic models and applied (high-dimensional) visual computing, learning from our visual perception.

The course is intended for brain scientists interested in functional modeling, and applied computer vision specialists interested in effective applications, and free for PhD students, staff, and interested members of Brainnetome, BMI and CAS Automation. Course notes and literature will be handed out. It will be a mix of physiology, medical imaging and computer vision.

Day 1, 09:00-12:00, 14:00-16:30 

1. Anatomy and physiology of the visual system: retina, thalamus, visual cortex, columnar pinwheels, feedback.

           Methods: Tractography, voltage sensitive dyes, calcium fluorescent intrinsic imaging, fMRI.

           Models in low-level, medium and high-level human vision and computer vision

2. Sell emergence of simple cell receptive fields in V1 for shape, motion and color differential structure

           Methods: first principles, axiomatics, principal component analysis

3. Multi-scale differential operators, image structure, a geometric model for simple and complex cells

          Applications: feature detection, deblurring Gaussian blur, curvatures and shape detection, focus: medical imaging

Day 2, 09:00-12:00, 14:00-16:30 

1. Modeling feedback mechanisms in the visual system, adaptive filtering.

           Application: edge-preserving smoothing,

2. Multi-scale image analysis, deep image structure, hierarchical models in vision

           Applications: edge focusing, SIFT landmark points, scene recognition

3. Multi-orientation pinwheel structures and multi-orientation contextual image analysis.

           Applications: brain tractography enhancement, optic radiation detection for safe epilepsy surgery

           Vessel tracking, retinal image analysis for diabetes screening

4. Outlook: geometric models, Lie groups and deep learning perspectives

Short biography: 

Prof. Bart M. ter Haar Romeny is professor at Eindhoven University of Technology in the Netherlands, Biomedical Image Analysis, Department of Biomedical Engineering. He received his MSc in Applied Physics from Delft University of Technology in 1978 and PhD from Utrecht University in 1983. He was Head of Physics at the Utrecht University Hospital Radiology Department and associate professor at the Image Sciences Institute (ISI) of Utrecht University (1989-2001).

He is distinguished professor at Northeastern University. His research interests focus on biologically inspired image analysis algorithms, multi-valued 3D visualization, especially brain connectivity, and computer-aided diagnosis (in particular for retinal image analysis for diabetes screening), and brain network analysis for epilepsy and neurosurgery. He is President of the Dutch Society for Pattern Recognition and Image Processing, and has been President of the Dutch Society for Biophysics & Biomedical Engineering (1998 – 2002) and the Dutch Society of Clinical Physics (NVKF, 1990-1992). He initiated the 'Scale-Space & Variational methods (SSVM)' conference series in 1997. He is reviewer for many journals and conferences, and organized several Summer Schools. He is an awarded teacher, and a frequent keynote lecturer. Prof.

Romeny is Senior Member of IEEE, Board member of IAPR, registered Clinical Physicist of NVKF, associate partner and visiting professor in the Chinese Brainnetome consortium, and recipient of the Mathematica Innovation Award and the Liaoning Friendship Award.