Imperial College London


Faculty of EngineeringDepartment of Electrical and Electronic Engineering




+44 (0)20 7594 6275j.barria Website




1012Electrical EngineeringSouth Kensington Campus






BibTex format

author = {Garcia-Trevino, E and Hameed, MZ and Barria, JA},
doi = {10.1145/3106369},
journal = {ACM Transactions on Knowledge Discovery From Data},
title = {Data stream evolution diagnosis using recursive wavelet densityestimators},
url = {},
volume = {12},
year = {2018}

RIS format (EndNote, RefMan)

AB - Data streams are a new class of data that is becoming pervasively important in a wide range of applications,ranging from sensor networks, environmental monitoring to finance.In this paper, we propose a novelframework for the online diagnosis of evolution of multidimensional streaming data which incorporatesRecursive Wavelet Density Estimators into the context of Velocity Density Estimation. In the proposedframework changes in streaming data are characterized by the use oflocalandglobal evolution coefficients.In addition, we propose for the analysis of changes in the correlation structure of the data a recursiveimplementation of the Pearson correlation coefficient using exponential discounting. Two visualisationtools, namely temporal and spatial velocity profiles, are extended inthe context of the proposed framework.Three are the main advantages of the proposed method over previous approaches: 1) the memory storagerequired is minimal and independent of any window size; 2) it has a significantly lower computationalcomplexity; and 3) it makes possible the fast diagnosis of data evolution atall dimensions and at relevantcombinations of dimensions with only one pass of the data. With the help offour examples, we show theframework’s relevance in a change detection context and its potential capability for real world applications.
AU - Garcia-Trevino,E
AU - Hameed,MZ
AU - Barria,JA
DO - 10.1145/3106369
PY - 2018///
SN - 1556-472X
TI - Data stream evolution diagnosis using recursive wavelet densityestimators
T2 - ACM Transactions on Knowledge Discovery From Data
UR -
UR -
VL - 12
ER -