INCORPORATION OF 3D NEURONAL I-V CHARACTERISTICS IN ARTIFICIAL NEURAL NETWORKS: I. ALGORITHM

ALAN W. L. CHIU¹*
¹Department of Biomedical Engineering, Louisiana Tech University, Ruston, LA, United States
* Corresponding Author : alanchiu@latech.edu

Received : 09-05-2011     Accepted : 02-06-2011     Published : 15-12-2011
Volume : 1     Issue : 2       Pages : 17 - 26
J Comput Simulat Model 1.2 (2011):17-26

The strategy of brain function replacement therapy is to restore the biological neuronal network by circumventing the damaged tissues with biomimetic neural models through innovative stimulation strategies. The computation capability of the artificial neural networks can be enhanced by incorporating neuronal characteristics. We propose a neural network paradigm using novel activation function that enables both sub-threshold linear dynamics and nonlinear supra-threshold spiking activities with hysteresis. Each neural unit consists of parallel linear (RC) components and nonlinear (activation) components. The overall aim of this paper entails the assembling and assessment of the aforementioned network. First, a novel 3D biologically-inspired activation function, obtained experimentally from rat hippocampal CA1 pyramidal neuron, is given. The activation function maps the current-voltage relationship between the ionic flow and electrical potential traversing through cell membrane. Second, an iterative learning rule is explained and illustrated by calculating the network’s synaptic weights as well as the scaling parameters in the activation function. Finally, the signals from a coupled linear system (mimicking sub-threshold activities driven by Gaussian white noise) are used to validate the learning rules. Our preliminary result suggests that the learning algorithm is able to obtain the appropriate synaptic weights and activation scaling factors in less than 10 iterations with mean square error of less than 0.01V.

[1] McCulloch W. and Pitts W.H. (1943) Bulletin of Mathematical Biophysics, 5, 115–133  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[2] Rumelhart D., Hinton G. and Williams R. (1986) Parallel Distributed Processing, 1, 318–362  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[3] Rieke F., Warland D., de Ruyter van Steveninck R. and Bialek W. (1997) Spikes: Exploring the Neural Code, MIT Press  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[4] Silberberg G., Bethge M., Markram H., Pawelzik K. and Tsodyks M. (2004) Journal of Neurophysiology, 91:704–709  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[5] Ferster D. and Spruston N. (1995) Science, 270: 756–757  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[6] Thorpe S., Delorme A. and Rullen R.V. (2001) Neural Networks, 14: 715–726  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[7] Huxter J., Burgess N. and O’Keefe, J. (2003) Nature, 425:828–832  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[8] Bohte S.M., Kok J.N. and Poutr´e H.L., (2000) Proceedings of the European Symposium on Artificial Neural Networks, 419–425  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[9] Christodoulou C., Bugmann G. and Clarkson T.G. (2002) Neural Networks, 15: 891–908  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[10] Geisler C. and Goldberg J.M. (1966) Biophysical Journal, 6: 53-69  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[11] Maass W. and Bishop C.M. (1999) Pulsed Neural Networks. MIT Press  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[12] Natschlager T., Maass W. and Zador A. (2001) Network: Computation in Neural Systems, 12: 75– 87  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[13] Mehrtash N., Jung D., Hellmich H.H., Schoenauer T., Lu V.T. and Klar H. (2003) IEEE Transactions on Neural Networks, 14: 980–992  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[14] Sekerli M., Negro C.A.D., Lee R.H. and Butera R.J. (2004) IEEE Transactions on Biomedical Engineering, 51:1665–1672  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[15] Swiercz W., Cios K.J., Staley K., Kurgen L., Accurso F. and Sagel S. (2006) IEEE Transactions on Neural Networks, 17:94–105  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[16] Haykin S. (2008) Neural networks and Learning Machines, Prentice Hall  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[17] Hodgkin L. and Huxley A.F. (1952) Journal of Physiology, 117:500-544  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[18] Fu P., Bardakjian B.L., D’Aguanno A. and Carlen P.L. (1989) IEEE Transactions on Biomedical Engineering, 36:55–64  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[19] Bardakjian B.L. (1989) IEEE 11th Annual Conference in EMBS, 11:1284–1285  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[20] Izhikevich E.M. (2003) IEEE Transactions on Neural Networks, 14:1569–1572  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[21] Rall W. (1969) Biophysical Journal, 9:1483–1508.  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[22] Rall W. (1977) Handbook of Physiology: the nervous system 1, American Physiological Society  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus  

[23] Marmarelis V.Z. (1994) Advanced Methods of Physiological System Modeling, Plenum Press  
» CrossRef   » Google Scholar   » PubMed   » DOAJ   » CAS   » Scopus