The Electronic Vision(s) Group designs micro chips capable of reproducing biologically realistic neural network models. A new chip, implementing multi-compartment models is currently in production and will be available soon for experiments. The announced thesis will reproduce theoretical results on this chip and work on transferring our single-compartment calibration algorithm onto the new framework. Background:
In vivo measurements of biological neural networks allow a macroscopic view on large networks with uncontrolled connectivity. In vitro experiments on the other hand can give a detailed view on single cell operation or small networks. Computer simulations are used to close the gap between both methods.
So called neuromorphic hardware tries to emulate the differential equations on the brain in a mostly analog fashion to construct microchips being able to reproduce biological behavior of the nervous system for extending knowledge about the brain and to gain access to its computational power. These devices can be an efficient alternative to computer simulations.
Most simulations and emulations use point-neuron models, which is a heavy simplification of a biological cell. The next more complex step is the use of a multi-compartment neuron model. Thesis:
We have developed a new micro chip with multi-compartment neurons which will be returned from production soon. We will provide first operation and basic experiments; more complex experiments will be done by the student.
A calibration algorithm exists for our single-compartment chips. This mechanism has to be transfered onto the new single-compartment neurons. Finally it has to be extended for multi-compartment calibration. Perspective:
Our projects are founded by the EU-project BrainScaleS which involves contacts to an international, interdisciplinary consortium of circa one hundred scientists. Several formal students of our group are now working in project partners' labs.
Neuromorphic Engineering is a growing field with expected major breakthroughs in the next years.
Practical knowledge gained in this thesis may be a good preparation for a job in the industry, too. Requirements:
- Basic C/CPP programming skills
- Interest in neuroscience and electronics
- Skills in Electronics (Lecture “Electronics for physicists” for example)