Large deviations properties of maximum entropy markov chains from spike trains

摘要: We consider the maximum entropy Markov chain inference approach to characterize collective statistics of neuronal spike trains, focusing on statistical properties inferred model. To find chain, we use thermodynamic formalism, which provides insightful connections with physics and thermodynamics from large deviations arise naturally. provide an accessible introduction problem theory community computational neuroscience, avoiding some technicalities while preserving core ideas intuitions. review techniques useful in train describe accuracy convergence terms sampling size. these results study fluctuation correlations, distinguishability, irreversibility chains. illustrate applications using simple examples where deviation rate function is explicitly obtained for models relevance this field.