Andrei Markov

Andrei Markov

A Russian mathematician who introduced what are now called Markov chains in 1906: sequences in which the probability of the next state depends only on the current one, not on the full history that preceded it. He demonstrated the idea in 1913 by counting vowel and consonant transitions across 20,000 letters of Pushkin's Eugene Onegin, an early piece of statistical text analysis. Markov chains underpin hidden Markov models, n-gram language models, Markov decision processes in reinforcement learning, and, more distantly, every model that predicts the next token in a sequence. (See also: Machine learning, Large language model, Tokenisation)