This tutorial will describe the process of fitting optimal parameters to single neuron models using parameter search techniques. Different methods will be compared with an emphasis on stochastic methods useful for ragged parameter landscapes. Different parameter searching methods, including genetic algorithms, simulated annealing, and conjugate gradient descent will be discussed using both simple models and models taken from research projects. Software tools, in particular the GENESIS "param" library, will be described in some detail Limitations of parameter searching techniques and future directions will also be considered.
Back to the index of tutorials