Using Genetic Algorithms to Optimize Neuron Models
Author Information
Author(s): Stock Raphael, Kaiser Jakob, Müller Eric, Schemmel Johannes, Schmitt Sebastian
Primary Institution: Heidelberg University
Hypothesis
Can genetic algorithms effectively parameterize multi-compartmental neuron models on neuromorphic hardware?
Conclusion
The study successfully demonstrated that genetic algorithms can replicate the desired behavior of excitatory postsynaptic potentials in neuron models.
Supporting Evidence
- The algorithm successfully replicated EPSP attenuation behavior in both single and multi-objective searches.
- Genetic algorithms were shown to be effective in parameterizing multi-compartmental neuron models.
- The study utilized a neuromorphic hardware platform to validate the results.
Takeaway
This study shows how computers can help us understand how brain cells work by finding the best settings for their models.
Methodology
Genetic algorithms were used to optimize neuron model parameters, validated through a grid search and spike-triggered averaging.
Limitations
The study's findings may be affected by the inherent trial-to-trial variations of the analog hardware.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website