Nonconvergence in Logistic and Poisson Models for Neural Spiking

Generalized linear models are an increasingly common approach for spike train data analysis. For the logistic and Poisson models, one possible difficulty is that iterative algorithms for computing parameter estimates may not converge because of certain data configurations. For the logistic model, these configurations are called complete and quasi-complete separation. We show that these features are likely to occur because of refractory periods of neurons. We use an example to study how standard software deals with this difficulty. For the Poisson model, we show that the same difficulties arise, this time possibly due to bursting or specifics of the binning. We characterize the nonconvergent configurations for both models, show that they can be detected by linear programming methods, and discuss possible remedies.

[1]  Robert E. Kass,et al.  A Spike-Train Probability Model , 2001, Neural Computation.

[2]  L. Paninski.  Maximum likelihood estimation of cascade point-process neural encoding models , 2004, Network.

[3]  G. Prasad,et al.  Logistic Regression Model with Surrogate Covariate , 2007 .

[4]  Micah Altman,et al.  Numerical Issues in Statistical Computing for the Social Scientist , 2003 .

[5]  R. Eubank.  Nonparametric Regression and Spline Smoothing , 1999 .

[6]  A. Albert,et al.  On the existence of maximum likelihood estimates in logistic regression models , 1984 .

[7]  John P. Cunningham,et al.  An L1-regularized logistic model for detecting short-term neuronal interactions , 2011, Journal of Computational Neuroscience.

[8]  Trevor Hastie,et al.  Regularization Paths for Generalized Linear Models via Coordinate Descent. , 2010, Journal of statistical software.

[9]  R. W. Wedderburn,et al.  On the existence and uniqueness of the maximum likelihood estimates for certain generalized linear models , 1976 .

[10]  M. Silvapulle.  On the Existence of Maximum Likelihood Estimators for the Binomial Response Models , 1981 .

[11]  Thomas J. Santner,et al.  A note on A. Albert and J. A. Anderson's conditions for the existence of maximum likelihood estimates in logistic regression models , 1986 .

[12]  J. F. C. Kingman,et al.  Information and Exponential Families in Statistical Theory , 1980 .

[13]  Micah Altman,et al.  Convergence Problems in Logistic Regression , 2004 .

[14]  David G. Luenberger,et al.  Linear and nonlinear programming , 1984 .

[15]  Ian H. Stevenson,et al.  Bayesian Inference of Functional Connectivity and Network Structure From Spikes , 2009, IEEE Transactions on Neural Systems and Rehabilitation Engineering.

[16]  R. Cook,et al.  Principal fitted components for dimension reduction in regression , 2008, 0906.3953.

[17]  Uri T Eden,et al.  A point process framework for relating neural spiking activity to spiking history, neural ensemble, and extrinsic covariate effects. , 2005, Journal of neurophysiology.

[18]  D. Brillinger.  Maximum likelihood analysis of spike trains of interacting nerve cells , 2004, Biological Cybernetics.

[19]  Rong Jin,et al.  Identifying Functional Connectivity in Large-Scale Neural Ensemble Recordings: A Multiscale Data Mining Approach , 2009, Neural Computation.