Markov Chain Data

The dataset is generated as random-walks on a graph with three nodes. At each time step, a signal is emitted as a random vector sampled from a 15-dimensional multivariate normal distribution with mean and covariance matrix specified by the node the random walk is at.