A hybrid method for novelty detection in time series based on states transitions and swarm intelligence

This paper introduces a novel instance-based one-class classification method for novelty detection in time series based on its states transition. The main feature of our work is to generate an efficient method which automatically finds the parameters (whose yields the best model) according with the quality of the discovered time series states and the validation error. This method involves clustering and reducing the number of samples in a training dataset which does not contain novelty samples. Experiments carried out using three real-world time series show that the proposed method is able to build models with a reduced number of stored prototypes. The results obtained by our method were compared with the results of the SAX and both methods have successfully detected the novelties, however, the parameters which resulted in the best SAX model were achieved without validation phase (i.e. analyzing the results obtained for the test set).

[1]  Philip Chan,et al.  Learning States and Rules for Detecting Anomalies in Time Series , 2005, Applied Intelligence.

[2]  Carsten Peterson,et al.  Finding the Embedding Dimension and Variable Dependencies in Time Series , 1994, Neural Computation.

[3]  James Kennedy,et al.  Particle swarm optimization , 2002, Proceedings of ICNN'95 - International Conference on Neural Networks.

[4]  Vipin Kumar,et al.  Chameleon: Hierarchical Clustering Using Dynamic Modeling , 1999, Computer.

[5]  Bilge Karaçali,et al.  Fast minimization of structural risk by nearest neighbor rule , 2003, IEEE Trans. Neural Networks.

[6]  Andries Petrus Engelbrecht,et al.  Fundamentals of Computational Swarm Intelligence , 2005 .

[7]  Eija Koskivaara.  Artificial Neural Networks in Auditing : State of the Art , 2003 .

[8]  Dipankar Dasgupta,et al.  Novelty detection in time series data using ideas from immunology , 1996 .

[9]  Tiago Alessandro Espínola Ferreira,et al.  A New Evolutionary Approach for Time Series Forecasting , 2007, 2007 IEEE Symposium on Computational Intelligence and Data Mining.

[10]  Eamonn J. Keogh,et al.  HOT SAX: efficiently finding the most unusual time series subsequence , 2005, Fifth IEEE International Conference on Data Mining (ICDM'05).

[11]  F. Takens.  Detecting strange attractors in turbulence , 1981 .

[12]  Adriano Lorena InÃcio de Oliveira,et al.  Neural networks forecasting and classification-based techniques for novelty detection in time series , 2004 .

[13]  N. Tanaka,et al.  Estimating the active demension of the dynamics in a time series based on an information criterion , 2001 .

[14]  J. Ma,et al.  Time-series novelty detection using one-class support vector machines , 2003, Proceedings of the International Joint Conference on Neural Networks, 2003..

[15]  Maurice Clerc,et al.  The particle swarm - explosion, stability, and convergence in a multidimensional complex space , 2002, IEEE Trans. Evol. Comput..

[16]  Jessica Lin,et al.  Visually mining and monitoring massive time series , 2004, KDD.

[17]  George Karypis,et al.  C HAMELEON : A Hierarchical Clustering Algorithm Using Dynamic Modeling , 1999 .

[18]  B. John Oommen,et al.  Enhancing prototype reduction schemes with recursion: a method applicable for "large" data sets , 2004, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics).

[19]  María Teresa Lozano Albalate,et al.  Data Reduction Techniques in Classification Processes , 2007 .

[20]  James Kennedy,et al.  Why does it need velocity? , 2005, Proceedings 2005 IEEE Swarm Intelligence Symposium, 2005. SIS 2005..