Likelihood-Based Approaches to Modeling the Neural Code
Jonathan W. Pillow · The MIT Press eBooks · 2006
This chapter discusses likelihood-based approaches to building mathematical models of the neural code. It introduces probabilistic neural models such as the linear-non-linear-Poisson (LNP) model (models of neural response), the generalized linear model (GLM), and the generalized integrate-and-fire (GIF) model. The chapter also examines the methods of evaluating the validity of probabilistic models, which includes cross-validation, time rescaling, and model-based decoding.