Stochastic Automata, Stochastic Languages, and Pattern Recognition

摘要

The paper describes some relations between stochastic automata and stochastic languages, and the possible application of stochastic languages to pattern recognition problems. Stochastic finite-state and stochastic context-free languages are defined. The problem of probability assignment in stochastic languages is discussed. The properties of stochastic finite automata and stochastic pushdown automata used as acceptors are investigated. The use of formal languages in pattern recognition problems is illustrated. The potential application of stochastic languages for pattern description is demonstrated, and the possibility of employing stochastic automata as pattern classifiers is discussed. (Author)