Stochastic training of a biologically plausible spino-neuromuscular system model

A primary goal of evolutionary robotics is to create systems that are as robust and adaptive as the human body. Moving toward this goal often involves training control systems that processes sensory information in a way similar to humans. Artificial neural networks have been an increasingly popular option for thisbecause they consist of processing units that approximate thesynaptic activity of biological signal processing units, i.e. neurons. In this paper we train a nonlinear recurrent spino-neuromuscular system model(SNMS) comparing the performance of genetic algorithms (GA)s, particle swarm optimizers (PSO)s, and GA/PSO hybrids. This model includes several key features of the SNMS that have previously been modeled individually but have not been combined into a single model as is done here. The results show that each algorithm produces fit solutions and generates fundamental biological behaviors that are not directly trained for such as tonic tension behaviors and tricepsactivation patterns.

[1]  A. E. Eiben,et al.  Introduction to Evolutionary Computing , 2003, Natural Computing Series.

[2]  Terence Soule,et al.  Breeding swarms: a GA/PSO hybrid , 2005, GECCO '05.

[3]  Wolfgang Maass,et al.  The Computational Power of Spiking Neurons Depends on the Shape of the Postsynaptic Potentials , 1996, Electron. Colloquium Comput. Complex..

[4]  Andrzej J. Kasinski,et al.  Experimental Demonstration of Learning Properties of a New Supervised Learning Method for the Spiking Neural Networks , 2005, ICANN.

[5]  Dario Floreano,et al.  Evolution of Spiking Neural Controllers for Autonomous Vision-Based Robots , 2001, EvoRobots.

[6]  V.P. Plagianakos,et al.  Spiking neural network training using evolutionary algorithms , 2005, Proceedings. 2005 IEEE International Joint Conference on Neural Networks, 2005..

[7]  S Cushing,et al.  Comparison of the morphological and electrotonic properties of Renshaw cells, Ia inhibitory interneurons, and motoneurons in the cat. , 2003, Journal of neurophysiology.

[8]  Günther Palm,et al.  Controlling the Speed of Synfire Chains , 1996, ICANN.

[9]  Bo Cartling,et al.  Control of computational dynamics of coupled integrate-and-fire neurons , 1997, Biological Cybernetics.

[10]  Michele Giugliano,et al.  Activity-Driven Computational Strategies of a Dynamically Regulated Integrate-and-Fire Model Neuron , 1999, Journal of Computational Neuroscience.

[11]  Nicolas Durand,et al.  Neural Nets Trained by Genetic Algorithms for Collision Avoidance , 2000, Applied Intelligence.

[12]  Michael Schmitt,et al.  Learning Temporally Encoded Patterns in Networks of Spiking Neurons , 2004, Neural Processing Letters.

[13]  X. Yao.  Evolving Artificial Neural Networks , 1999 .

[14]  Thomas A. McMahon,et al.  Muscles, Reflexes, and Locomotion , 1984 .

[15]  M. Häusser,et al.  Neurobiology , 2001, Current Opinion in Neurobiology.

[16]  Ehsan Arabzadeh,et al.  New attractor states for synchronous activity in synfire chains with excitatory and inhibitory coupling , 2002, Biological Cybernetics.

[17]  Louis Tao,et al.  Efficient and Accurate Time-Stepping Schemes for Integrate-and-Fire Neuronal Networks , 2001, Journal of Computational Neuroscience.

[18]  M. Mandischer.  Evolving recurrent neural networks with non-binary encoding , 1995, Proceedings of 1995 IEEE International Conference on Evolutionary Computation.

[19]  Daniel Lehmann,et al.  Modeling Compositionality by Dynamic Binding of Synfire Chains , 2004, Journal of Computational Neuroscience.

[20]  Kenneth Meijer,et al.  Muscle contraction history: modified Hill versus an exponential decay model , 2000, Biological Cybernetics.

[21]  Russell C. Eberhart,et al.  A discrete binary version of the particle swarm algorithm , 1997, 1997 IEEE International Conference on Systems, Man, and Cybernetics. Computational Cybernetics and Simulation.

[22]  Josh E. Baker,et al.  A thermodynamic muscle model and a chemical basis for A.V. Hill's muscle equation , 2000, Journal of Muscle Research & Cell Motility.