Proceedings of the 32nd conference on Winter simulation
Spiking Neuron Models: An Introduction
Spiking Neuron Models: An Introduction
Convergence Properties of the Nelder--Mead Simplex Method in Low Dimensions
SIAM Journal on Optimization
Neural Computation
A Method for Selecting the Bin Size of a Time Histogram
Neural Computation
Fast Gaussian process methods for point process intensity estimation
Proceedings of the 25th international conference on Machine learning
Kernel bandwidth optimization in spike rate estimation
Journal of Computational Neuroscience
| Hi-index | 0.01 |
Firing rate metrics are often used by researchers to evaluate the information encoded in neural responses based on the rate at which one or more neurons generate spikes. Since the true firing rate of a set of neurons is unknown, it must be estimated from the neural responses. This paper analyzes the performance of the firing rate estimators and their impact on a typical neural code metric, the firing rate mean squared error. The provided analysis shows that the smoothing introduced by firing rate estimators must take into consideration the number of spike trains used in the estimation process. Furthermore, it unveils that these estimators can introduce significant bias into the neural metrics' results, which is of great importance when the objective is to tune the parameters of a model. To avoid the introduction of bias, an improved metric is proposed to be used as the objective function for neural model parameter estimation. This paper shows that the proposed metric can significantly improve the estimation of the model parameters, based both on theoretical and experimental results, with real and simulated data.