Neuromorphic Sensors and Processing

Members: Bragi Lovetrue, Fabien Alibart, Amir Reza, Antonio Ríos, Aristeidis Tsitiridis, Christopher Bennett, Christoph Posch, Damir Vodenicarevic, Dan Neil, Nasim Farahini, Francesco Galluppi, Jorg Conradt, Julien Martel, Jordi-Ysard Puigbó Llobet, Leon Bonde Larsen, Lukas Everding, Marc Osswald, Marcel Stimberg, Matthew Tata, Orhan Celiker, Ole Richter, Paul B Isaac's, Michael Pfeiffer, Yulia Sandamirskaya, Scott Stone, Shih-Chii Liu, Sim Bamford, Siohoi Ieng, Subhrajit Roy, Tobi Delbruck, Gianvito Urgese, Viviane Ghaderi, Wenjia Meng

Coordinators: Tobi Delbruck, Shih-Chii Liu, Ryad Benosman, Marc Osswald, Bernabe Linares-Barranco, Christoph Posch and others welcome

This workgroup will coordinate activities in processing visual and auditory neuromorphic sensor outputs.


  1. Meet and discuss projects
  2. Discuss additional projects and demonstrate jAER in action and help people install it
  3. Discuss projects and show how to write a basic event filter in jAER using netbeans
  4. Discuss example strategies for filtering, tracking and labeling events in jAER


  • Labyrinth Game - Tobi Delbruck. See YouTube video for past incarnation
  • Learning an optimum Labyrinth ball controller - Danny Neil
  • Playing with DAVIS and Occulus Rift (DAVIS user guide, DAVIS camera brief)
  • Vocal Sensorimotor Learning - Damip, Leon, Christopher
  • Using the DAS for digit recognition - Subhrajit
  • Making a new DisplayMethod for jAER to show spike events in a rolling 3D space time view. Learn to use display lists and maybe an OpenGL shader program to build a cool new display method that shows the dynamic nature of the data - Tobi Delbruck
  • Record DAVIS data while driving around the neighborhood, and then label the road center using the <i>TargetLabeler?</i> and try to use a deep network to identify the road - Tobi Delbruck
  • Work on a face-detector CNN using recorded DAVIS sensor output
  • Improve the UDP interface from jAER to SpiNNaker


Project Name : Using the DAS for digit recognition

Participants: Subhrajit Roy

Problem: In this project we tried to classify the spatio-temporal spike trains obtained from the DAS chip when the spoken digits of the TI46 database are presented to it.

Idea: The idea was to use structural plasticity based learning for neurons with non-linear dendrites and binary synapses which we developed in NTU last year. The structure, known as Dendritically Enhanced Readout (DER), is shown in the figure which consists of two neuronal cells with each cell having m dendrites. Each dendrite has k synaptic connections of weight 1 from the input. The learning rule involves changes in connections in between the input lines and the synapses over epochs.

DAS spike data applied to DER

Results: Currently the DER readout is capable of only doing two-class classification. So we selected spike data for only two digits (4 and 5 – according to DAS researchers these two are the most difficult to classify correctly) and trained DER with them. We got a training accuracy of 93.6% and a testing accuracy of 91.4%. The accuracies have been averaged over 10 trials. After the workshop, the first thing we would like to do is a multi-class expansion of DER for training over the entire DAS spike dat corresponding to TI46 dataset.

Last modified 4 years ago Last modified on 05/14/15 10:30:19

Attachments (1)

Download all attachments as: .zip