Fig. 4: Individual Distribution of Latent states.

a Distribution of z⁽³⁾μ post-training, as shown in Fig. 2b, overlaid with an estimated probability density plot. Each layer of varying shades represents boundaries of corresponding probability levels. Estimation of bivariate probability density utilized the kernel density estimator from the “ggdensity” package in R. b–d Distributions of z⁽³⁾μ depicted in Fig. 3a are shown separately for each of three individuals in the training data. e Shows the distribution of z⁽³⁾μ estimations from data assimilation applied to ECoG sequences from a test individual.