Recurrence Plots and Cross Recurrence Plots

Proceedings of the Special Session of the Seventh Mexican International Conference on Artificial Intelligence - MICAI 2008, 111–116p. (2008) DOI:10.1109/MICAI.2008.16

Classification of Data Sequences by Similarity Analysis of Recurrence Plot Patterns

Quantification of the similarity between data sequences is important in different database and data mining tasks such as indexing, retrieving, clustering and classification, many similarity metrics (Euclidean, DTW, among others) operate directly in the raw representation of the sequences, but this implies when the sequences are compared, not take into account information about the collective behavior of the data that forms a sequence and the hidden relations between such data, such information can be important for classification of sequences based on their structures and their relations with the dynamics that such structures can represent (e.g. stationary, random, complex). We propose a computational technique for similarity analysis and classification of recurrence plot patterns: RecurrenceVs. The results show that the proposed technique is able to classify data sequences by similarity families based on the recurrence plot patterns, which preserve the information about the structure and dynamics represented by the data sequences.