Statistical analysis of neural data: The expectation-maximization (EM) algorithm, mixture models, and applications to spike sorting
چکیده
0.1 Bound optimization; auxiliary functions . . . . . . . . . . . . . . . . . . . . . 2 0.2 The EM algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 0.3 EM may be used to optimize the log-posterior instead of the log-likelihood . . 5 0.4 Example: Mixture models and spike sorting . . . . . . . . . . . . . . . . . . . 5 0.5 Example: Spike sorting given stimulus observations . . . . . . . . . . . . . . . 7 0.6 Example: LNP models with spike-timing jitter . . . . . . . . . . . . . . . . . 8 0.7 Fitting hierarchical generalized linear models for spike trains . . . . . . . . . 11 0.8 Connection between EM algorithm and computing gradients of the marginal likelihood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13