Markov Chains: The Machine ( I )
Paul Aurelian Gagniuc · 2021
This chapter presents a series of discussions about the methodology involved in the computation of transition probabilities and the transition matrices by which these numerical values are represented. It describes the main steps by which the discrete probability detector (DPD) algorithm is able to transform any sequence of symbols into a first-order transition matrix. The transition matrices produced by the DPD algorithm are used by a Markov chains generator (MCG) for predictions. The original DPD algorithm is organized in four phases: alphabet detection, matrix initialization, frequency detection, and calculation of transition probabilities. A transition matrix is calculated based on a training sequence. The output of an MCG mimics the training sequence and the process itself represents a prediction. The MCG represented a machine mimicking/predicting the output of the system that generated the sequence of observations initially used in the construction of the transition matrix.