ICF Iterative Cross-Correlation Filter

ICF – An Iterative Cross-Correlation Filter for Sensor Time series

The software in this page provides the Matlab implementation of two unsupervised feature subset selection algorithms specifically tailored to sensor time series data. A brief description of the two algorithms follows: to dowload the code click on the Matlab icon.

matlab  The ICF (Iterative Cross-correlation Filter) algorithm has been developed in the context of the EU FP7 RUBICON project and it has been originally published in [1] and later extended in [2]. ICF implements a forward selection-elimination procedure (without backtracking) that targets the reduction of feeature redundancy measured by Normalized Cross-Correlation.
For further details on the ICF algorithm see [1] and [2], as well as the associated slides presented @ EANN 2014.

matlab The CleVer method  is an efficient feature filter algorithm for multivariate timeseries presented in [3]. It exploits the properties of the principal components common to all the timeseries to provide a ranking of the more informative features. Several versions of the CleVer algorithm exists differing for the approach used to rank the variables with respect ot the common principal components. The code implements the CleVer-Cluster approach exploiting the k-means clustering algorithm for variable ranking (see details in [3]). The CleVer algorithm has been used as a baseline to assess the performance of ICF in [1] [2].

The code is provided as is with no warranty and technical support. Please inform the author (Davide Bacciu) if you intend to redistribute the code.

[1] D. Bacciu, “An Iterative Feature Filter for Sensor Timeseries in Pervasive Computing Applications”,  Proceedings of the 15th Conference on Engineering Applications of Neural Networks (EANN’14), Communications in Computer and Information Science, Vol. 459, pp. 29-48, Springer International Publishing, 2014

[2] D. Bacciu, “Unsupervised feature selection for sensor time-series in pervasive computing applications.” Neural Computing and Applications: 1-15, Springer, 2015 (Online)

[3] H. Yoon, K. Yang, C. Shahabi “Feature subset seletion and feature ranking for multivariate time series”, IEEE Transactions on Knowledge and Data Engineering, Vol. 17(9), pp. 1186-1198, IEEE, 2005

Citation

If you find this code useful, please remember to cite:

Bacciu Davide: Unsupervised feature selection for sensor time-series in pervasive computing applications. In: Neural Computing and Applications, 27 (5), pp. 1077-1091, 2016, ISSN: 1433-3058.

BibTeX (Download)

@article{icfNca15,
title = {Unsupervised feature selection for sensor time-series in pervasive computing applications},
author = {Bacciu Davide},
url = {http://pages.di.unipi.it/bacciu/wp-content/uploads/sites/12/2016/04/nca2015.pdf},
doi = {10.1007/s00521-015-1924-x},
issn = {1433-3058},
year  = {2016},
date = {2016-07-01},
journal = {Neural Computing and Applications},
volume = {27},
number = {5},
pages = {1077-1091},
publisher = {Springer London},
abstract = {The paper introduces an efficient feature selection approach for multivariate time-series of heterogeneous sensor data within a pervasive computing scenario. An iterative filtering procedure is devised to reduce information redundancy measured in terms of time-series cross-correlation. The algorithm is capable of identifying nonredundant sensor sources in an unsupervised fashion even in presence of a large proportion of noisy features. In particular, the proposed feature selection process does not require expert intervention to determine the number of selected features, which is a key advancement with respect to time-series filters in the literature. The characteristic of the prosed algorithm allows enriching learning systems, in pervasive computing applications, with a fully automatized feature selection mechanism which can be triggered and performed at run time during system operation. A comparative experimental analysis on real-world data from three pervasive computing applications is provided, showing that the algorithm addresses major limitations of unsupervised filters in the literature when dealing with sensor time-series. Specifically, it is presented an assessment both in terms of reduction of time-series redundancy and in terms of preservation of informative features with respect to associated supervised learning tasks.},
keywords = {ambient assisted living, Echo state networks, feature selection, multivariate time-series, pervasive computing, structured data processing, wireless sensor networks},
pubstate = {published},
tppubtype = {article}
}
Bacciu Davide: An Iterative Feature Filter for Sensor Timeseries in Pervasive Computing Applications. Communications in Computer and Information ScienceEngineering Applications of Neural Networks, 459 , Springer International Publishing, 2014.

BibTeX (Download)

@conference{icfEann14,
title = {An Iterative Feature Filter for Sensor Timeseries in Pervasive Computing Applications},
author = {Bacciu Davide},
doi = {10.1007/978-3-319-11071-4_4},
year  = {2014},
date = {2014-01-01},
booktitle = {Communications in Computer and Information ScienceEngineering Applications of Neural Networks},
journal = {COMMUNICATIONS IN COMPUTER AND INFORMATION SCIENCE},
volume = {459},
pages = {39--48},
publisher = {Springer International Publishing},
abstract = {The paper discusses an efficient feature selection approach for multivariate timeseries of heterogeneous sensor data within a pervasive computing scenario.  An iterative filtering procedure is devised
to reduce information redundancy measured in terms of timeseries cross-correlation. The algorithm is capable of identifying non-redundant sensor sources in an unsupervised fashion even in presence of a large proportion of noisy features. A comparative experimental analysis on real-world data
from pervasive computing applications is provided, showing that the algorithm addresses major limitations of unsupervised filters in literature when dealing with sensor timeseries.},
keywords = {ambient assisted living, feature selection, multivariate time-series, pervasive computing, structured data processing, unsupervised learning, wireless sensor networks},
pubstate = {published},
tppubtype = {conference}
}
Top