Search for: [Abstract = "In this paper we developed a technique of an envelopes approximation of time series, and an application of the envelopes for solving a classification problem of time series. Next, we introduced upper and lower envelopes as well as a way to compress them by a neural network. Obtained in such a way essential attributes were used to generate rules for time series classification. The obtained rules can classify the testing example with 100 % accuracy."]

Number of results: 1