An Evolutionary Framework for Replicating Neurophysiological Data with Spiking Neural Networks
Here we present a framework for the automatic tuning of spiking neural networks (SNNs) that utilizes an evolutionary algorithm featuring indirect encoding to achieve a drastic reduction in the dimensionality of the parameter space, combined with a GPU-accelerated SNN simulator that results in a considerable decrease in the time needed for fitness evaluation, despite the need for both a training and a testing phase. We tuned the parameters governing a learning rule called spike-timing-dependent plasticity (STDP), which was used to alter the synaptic weights of the network. We validated this framework by applying it to a case study in which synthetic neuronal firing rates were matched to electrophysiologically recorded neuronal firing rates in order to evolve network functionality. Our framework was not only able to match their firing rates, but also captured functional and behavioral aspects of the biological neuronal population, in roughly 50 generations.
KeywordsSpiking neural networks Evolutionary algorithms Indirect encoding Neurophysiological recordings Plasticity Data matching Parallel computing
Supported by the National Science Foundation (Award IIS-1302125).
- 2.Asher, D.E., Krichmar, J.L., Oros, N.: Evolution of biologically plausible neural networks performing a visually guided reaching task. In: Proceedings of the 2014 Annual Conference on Genetic and Evolutionary Computation (GECCO 2014), pp. 145–152. ACM, New York (2014)Google Scholar
- 3.Beyeler, M., Carlson, K.D., Chou, T.-S., Dutt, N., Krichmar, J.L.: CARLsim 3: a user-friendly and highly optimized library for thecreation of neurobiologically detailed spiking neural networks. In: 2015 International Joint Conference on Neural Networks (IJCNN 2015), pp. 1–8. IEEE (2015)Google Scholar
- 4.Bi, G.-Q., Poo, M.-M.: Synaptic modifications in cultured hippocampal neurons: dependence on spike timing, synaptic strength, and postsynaptic cell type. J. Neurosci. 18(24), 10464–10472 (1998)Google Scholar
- 5.Carlson, K.D., Nageswaran, J.M., Dutt, N., Krichmar, J.L.: An efficient automated parameter tuning framework for spiking neuralnetworks. Front. Neurosci. 8 (2014)Google Scholar
- 6.Carlson, K.D., Richert, M., Dutt, N., Krichmar, J.L.: Biologically plausible models of homeostasis and STDP: stabilityand learning in spiking neural networks. In: The 2013 International Joint Conference on Neural Networks (IJCNN 2013), pp. 1–8. IEEE (2013)Google Scholar
- 8.Fountas, Z., Shanahan, M.: GPU-based fast parameter optimization for phenomenological spikingneural models. In: 2015 International Joint Conference on Neural Networks (IJCNN 2015), pp. 1–8, July 2015Google Scholar
- 12.Krichmar, J.L., Coussy, P., Dutt, N.: Large-scale spiking neural networks using neuromorphic hardwarecompatible models. ACM J. Emerging Technol. Comput. Syst. (JETC), 11(4) (2015). Article no. 36Google Scholar
- 16.Rossant, C., Goodman, D.F.M., Fontaine, B., Platkiewicz, J., Magnusson, A.K., Brette, R.: Fitting neuron models to spike trains. Front. Neurosci. 5(9) (2011)Google Scholar
- 20.Tripathy, S.J., Savitskaya, J., Burton, S.D., Urban, N.N., Gerkin, R.C.: Neuroelectro: a window to the world’s neuron electrophysiology data. Front. Neuroinf. 8 (2014)Google Scholar