ONLINE ANOMALY DETECTION WITH CONSTANT FALSE ALARM RATE

Date

2015-09-20

Author

Ozkan, Huseyin
Ozkan, Fatih
Delibalta, Ibrahim
KOZAT, SÜLEYMAN SERDAR

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

169
views

0
downloads

We propose a computationally highly scalable online anomaly detection algorithm for time series, which achieves - with no parameter tuning- a specified false alarm rate while minimizing the miss rate. The proposed algorithm sequentially operates on a fast streaming temporal data, extracts the nominal attributes under possibly varying Markov statistics and then declares an anomaly when the observations are statistically sufficiently deviant. Regardless of whether the source is stationary or non-stationary, our algorithm is guaranteed to closely achieve the desired false alarm rates at negligible computational costs. In this regard, the proposed algorithm is highly novel and appropriate especially for big data applications. Through the presented simulations,we demonstrate that our algorithm outperforms its competitor, i.e., the Neyman-Pearson test that relies on the Monte Carlo trials, even in the case of strong non-stationarity.

Subject Keywords

Anomaly Detection, Online, Efficient, Neyman-Pearson, NP, False alarm, Type-I error, Time Series

URI

https://hdl.handle.net/11511/67471

Conference Name

IEEE International Workshop on Machine Learning for Signal Processing

Collections

Graduate School of Natural and Applied Sciences, Conference / Seminar

Suggestions

OpenMETU
Core

Online Anomaly Detection Under Markov Statistics With Controllable Type-I Error Ozkan, Huseyin; Ozkan, Fatih; KOZAT, SÜLEYMAN SERDAR (2016-03-15) We study anomaly detection for fast streaming temporal data with real time Type-I error, i. e., false alarm rate, controllability; and propose a computationally highly efficient online algorithm, which closely achieves a specified false alarm rate while maximizing the detection power. Regardless of whether the source is stationary or nonstationary, the proposed algorithm sequentially receives a time series and learns the nominal attributes-in the online setting-under possibly varying Markov statistics. Then...
Consensus clustering of time series data Yetere Kurşun, Ayça; Batmaz, İnci; İyigün, Cem; Department of Scientific Computing (2014) In this study, we aim to develop a methodology that merges Dynamic Time Warping (DTW) and consensus clustering in a single algorithm. Mostly used time series distance measures require data to be of the same length and measure the distance between time series data mostly depends on the similarity of each coinciding data pair in time. DTW is a relatively new measure used to compare two time dependent sequences which may be out of phase or may not have the same lengths or frequencies. DTW aligns two time serie...
Online calibration of strapdown magnetometers Yigitler, Hüseyin; Leblebicioğlu, Mehmet Kemal (2009-01-01) In this work, an online calibration algorithm for strapdown magnetometers is proposed. The proposed method is based on the observation that the magnetometer's measurements are on an ellipsoid manifold. The algorithm is attitude independent and makes use of Extended Kalman Filter.
Rigorous optimizations of three dimensional antenna arrays using full wave simulations Onol, Can; Gokce, Ozer; Boyacı, Huseyın; Ergül, Özgür Salih (null; 2015-07-09) We present optimizations of three-dimensional antenna arrays using heuristic techniques coupled with the multilevel fast multipole algorithm (MLFMA). Without resorting to any periodicity and infinity assumptions, antenna arrays are modeled with surface integral equations and simulated via MLFMA, which also enables the analysis of arrays with non-identical elements. Genetic algorithms and particle swarm optimization methods are employed on the complex data produced by MLFMA in phasor domain to find optimal s...
Upper bounds to error probability with feedback Nakiboğlu, Barış (2010-01-22) A new analysis technique is suggested for bounding the error probability of fixed length block codes with feedback on discrete memoryless channels from above. Error analysis is inspired by Gal lager's error analysis for block codes without feedback. Using Burnashev-Zigangirov-D'yachkov encoding scheme analysis recovers previously known best results on binary symmetric channels and improves up on the previously known best results on k-ary symmetric channels and binary input channels.

Citation Formats

H. Ozkan, F. Ozkan, I. Delibalta, and S. S. KOZAT, “ONLINE ANOMALY DETECTION WITH CONSTANT FALSE ALARM RATE,” Boston, NY, 2015, p. 0, Accessed: 00, 2020. [Online]. Available: https://hdl.handle.net/11511/67471.