In this paper, some properties of stochastic context-free programmed grammars arepresented. By illustrative examples, the class of stochastic context-free programmed grammars is shown to be quite effective in characterizing the structures of some pictorial patterns. Two stochastic syntactic-analysis algorithms for stochastic context-free programmed grammars are introduced. It is shown that the stochastic analyzer with a fixed strategy would require the least average number of steps to recognize the language generated by a stochastic context-free programmed grammar, while the stochastic syntactic analyzer with a random strategy would be useful in the case where the probability information associated with the stochastic context-free programmed grammar is incomplete or inaccurately known. The performance of the stochastic syntactic analyzer with a random strategy would be better than that of the nondeterministic syntactic analyzer proposed earlier. The use of the proposed stochastic syntactic-analysis algorithm with a fixed strategy for syntactic pattern recognition is illustrated by a chromosome patternclassification experiment.
ASJC Scopus subject areas
- Environmental Science(all)
- Earth and Planetary Sciences(all)