The Hierarchical ICI-based Change Detection Test

The Problem

Intelligent systems meant to operate in nonstationary environments have to detect possible changes to proactively interact with the environment and adapt to evolving non-stationary conditions. One of the most appealing solutions consists in monitoring the statistical behavior of data by means of nonparametric sequential Change-Detection Tests (CDTs). These tests assess the stationarity of the data-generating process without requiring any a-priori information about the process or change.

One of the main drawbacks of CDTs are false positives, i.e., detections not corresponding to an actual change in the data-generating process, as these induce intelligent systems to raise false alarms.

The Hierarchical ICI-based CDT

The hierarchical ICI-based CDT [Alippi et al. 2011 a] is a nonparametric CDT characterized by two-levels:

  • The Detection Layer, which operates on-line and performs a sequential analysis of the observation stream. The purpose of this layer is to provide a prompt detection of possible nonstationarities; at the detection layer the ICI-based CDT is enforced [Alippi et al. 2011 b].
  • The Validation Layer, which is activated after a detection raised by the detection layer. The purpose of this layer is to validate the prospective nonstationarity by means of an hypothesis test on the features of the ICI-based CDT.

Such a hierarchical CDT significantly reduces the number of false positives and the detection delays, since the validation layer allows for tuning the ICI-based CDT to provide prompter detections. A distributed solution meant for wireless sensor networks has been presented in [Alippi et al. 2011 c].


The scheme of the hierarchical ICI-based CDT


The hierarchical ICI-based CDT belongs to the family of ICI-based CDTs [Alippi et al. 2011 b], which extract a set of features from data and assess the stationarity of these features by means of the ICI rule [Goldenshluger and Nemirovski 1997]. For a detailed description of the hierarchical ICI-based CDT, please refer to [Alippi et al. 2011 a].


Codes are available for download

A Matlab package contains both functions implementing the hierarchical ICI-based CDT and as dataset generation. A script (main.m) illustrates the basic usage of this Matlab package.

Download the Matlab Package




[Alippi et al. 2011 a] A Hierarchical, Nonparametric Sequential Change-Detection Test
Cesare Alippi, Giacomo Boracchi and Manuel Roveri, in Proceedings of IJCNN 2011, the International Joint Conference on Neural Networks, San Jose, California July 31 - August 5, 2011. pp 2889 - 2896,

[Alippi et al. 2011 b] A just-in-time adaptive classification system based on the intersection of confidence intervals rule, Cesare Alippi, Giacomo Boracchi, Manuel Roveri Neural Networks, Elsevier vol. 24 (2011), pp. 791-800

[Alippi et al. 2011 c] A distributed Self-adaptive Nonparametric Change-Detection Test for Sensor/Actuator Networks, Cesare Alippi, Giacomo Boracchi and Manuel Roveri, ICANN 2011, 21th International Conference on Artificial Neural Networks, June 14-17th, 2011, Espoo, Finland. Lecture Notes in Computer Science, 2011, Vol. 6792/2011, 173-180,

[Goldenshluger and Nemirovski 1997] On spatial adaptive estimation of nonparametric regression, Goldenshluger, A., & Nemirovski, A. (1997) Mathematical Methods of Statistics, vol 6, 135-170




All Matlab codes on this page are published under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License.

Creative Commons License