Detecting rare events using extreme value statistics applied to epileptic convulsions in children

Luca, Stijn; Karsmakers, Peter; Cuppens, Kris; Croonenborghs, Tom; Vel, Anouk Van de; Ceulemans, Berten; Lagae, Lieven; Huffel, Sabine Van; Vanrumste, Bart

doi:10.1016/j.artmed.2013.11.007

Cited by 33 publications

(29 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The method is evaluated using as well simulated as real world data and compared with an one-class support vector machine (SVM) approach. For the real world data this will lead to an increase up to 12.9% of the positive predictive values while the sensitivity scores stay unaltered in comparison with the results obtained in [1].…”

Section: B Contributionsmentioning

confidence: 65%

See 1 more Smart Citation

Anomaly Detection Using the Poisson Process Limit for Extremes

Luca

Karsmakers

Vanrumste

2014

2014 IEEE International Conference on Data Mining

Self Cite

View full text Add to dashboard Cite

Anomaly detection starts from a model of normal behavior and classifies departures from this model as anomalies. This paper introduces a statistical non-parametric approach for anomaly detection that is based on a multivariate extension of the Poisson point process model for univariate extremes. The method is demonstrated on both a synthetic and a real-world data set, the latter being an unbalanced data set of acceleration data collected from movements of 7 pediatric patients suffering from epilepsy that is previously studied in [1]. The positive predictive values could be improved with an increase up to 12.9% (and a mean of 7%) while the sensitivity scores stayed unaltered. The proposed method was also shown to outperform an one-class SVM classifier. Because the Poisson point process model of extremes is able to combine information on the number of excesses over a fixed threshold with that on the excess values, a powerful model to detect anomalies is obtained that can be of high value in many application.

show abstract

Section: B Contributionsmentioning

confidence: 65%

“…[3]), healthcare (e.g. [4] and [1]). All these applications have in common that data describing failure conditions or abnormal behavior are rare.…”

Section: Introductionmentioning

confidence: 99%

Anomaly Detection Using the Poisson Process Limit for Extremes

Luca

Karsmakers

Vanrumste

2014

2014 IEEE International Conference on Data Mining

Self Cite

View full text Add to dashboard Cite

show abstract

“…Anomaly detection using EVT approach is based on a model of normal behavior which was presented by the probability distribution function. In this section we borrow the methodology which was proposed in [3].…”

Section: A Methods 1 -Extreme Value For Measuring Anomaly Score (Evmas)mentioning

confidence: 99%

“…In the abnormal usage detection based on sequence patterns, there have been many proposed methods for detection abnormal usages of appliances in a smart home environment, such as in [3][4][5]. Moreover, Neural network approach has been done to predict the future values which are used to inform the caregiver in case anomalous behavior is predicted [6,7].…”

Section: Related Workmentioning

confidence: 99%

“…Previous researches of abnormal detection mainly focused on sequence pattern [6] and probability density function based on EVT [3,[8][9][10]. Furthermore, a heuristic novelty threshold has set on the pdf f(x)=k, such that x is abnormal when f(x)<k. f(x) is used simply as abnormal score, and the threshold is set such that separation between normal and abnormal data is maximized on the validation dataset [11].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Methods for Abnormal Usage Detection in Developing Intelligent Systems for Smart Homes

Minh-Thai

Thai-Nghe

2015

2015 Seventh International Conference on Knowledge and Systems Engineering (KSE)

View full text Add to dashboard Cite

Detecting abnormal usages in intelligent systems, especially in smart home systems, is an important task. By exploring log data, useful information and patterns can be discovered which may help users/organizations to better understand the usage of their appliances and to distinguish unnecessary usages as well as abnormal problems which can cause waste, damages, or even fire. This work proposes several methods which can be used to detect the abnormal usages of appliances in smart home system and proposes a parameter tuning strategy to optimize the model's accuracy. The proposed methods are validated by using a real dataset. Experimental results show that these approaches perform nicely and could be applied for practice.

show abstract