Change Point Detection Method Research Articles

Tail-greedy bottom-up data decompositions and fast multiple change-point detection

This article proposes a “tail-greedy”, bottom-up transform for one-dimensional data, which results in a nonlinear but conditionally orthonormal, multiscale decomposition of the data with respect to an adaptively chosen unbalanced Haar wavelet basis. The “tail-greediness” of the decomposition algorithm, whereby multiple greedy steps are taken in a single pass through the data, both enables fast computation and makes the algorithm applicable in the problem of consistent estimation of the number and locations of multiple change-points in data. The resulting agglomerative change-point detection method avoids the disadvantages of the classical divisive binary segmentation, and offers very good practical performance. It is implemented in the R package breakfast, available from CRAN.

FreSpeD: Frequency-Specific Change-Point Detection in Epileptic Seizure Multi-Channel EEG Data

ABSTRACTThe goal in this article is to develop a practical tool that identifies changes in the brain activity as recorded in electroencephalograms (EEG). Our method is devised to detect possibly subtle disruptions in normal brain functioning that precede the onset of an epileptic seizure. Moreover, it is able to capture the evolution of seizure spread from one region (or channel) to another. The proposed frequency-specific change-point detection method (FreSpeD) deploys a cumulative sum-type test statistic within a binary segmentation algorithm. We demonstrate the theoretical properties of FreSpeD and show its robustness to parameter choice and advantages against two competing methods. Furthermore, the FreSpeD method produces directly interpretable output. When applied to epileptic seizure EEG data, FreSpeD identifies the correct brain region as the focal point of seizure and the timing of the seizure onset. Moreover, FreSpeD detects changes in cross-coherence immediately before seizure onset which indicate an evolution leading up to the seizure. These changes are subtle and were not captured by the methods that previously analyzed the same EEG data. Supplementary materials for this article are available online.

An ensemble-based change-point detection method for identifying unexpected behaviour of railway tunnel infrastructures

A large amount of data is generated by Structural Health Monitoring (SHM) systems and, as a consequence, processing and interpreting this data can be difficult and time consuming. Particularly, if work activities such as maintenance or modernization are carried out on a bridge or tunnel infrastructure, a robust data analysis is needed, in order to accurately and quickly process the data and provide reliable information to decision makers. In this way the service disruption can be minimized and the safety of the asset and the workforce guaranteed.In this paper a data mining method for detecting critical behaviour of a railway tunnel is presented. The method starts with a pre-processing step that aims to remove the noise in the recorded data. A feature definition and selection step is then performed to identify the most critical area of the tunnel. An ensemble of change-point detection algorithms is proposed, in order to analyse the critical area of the tunnel and point out the time when unexpected behaviour occurs, as well as its duration and location. The work activities, which are carried out at the time of occurrence of the critical behaviour and have caused this behaviour, are finally identified from a database of the work schedule and used for the validation of the results. Using the proposed method, fast and reliable information about infrastructure condition is provided to decision makers.

Change-point detection method for clinical decision support system rule monitoring.

A clinical decision support system (CDSS) helps clinicians to manage patients, but malfunctions of its components or other systems on which it depends may affect its intended functions. Monitoring the system and detecting changes in its behavior that may indicate the malfunction can help to avoid any potential costs associated with its improper operation. In this paper, we investigate the problem of detecting changes in the CDSS operation, in particular its monitoring and alerting subsystem, by monitoring its rule firing counts. We aim to screen and detect changes in real-time, that is whenever a new datum (rule firing count) arrives, we want to have a score indicating how likely there is a change in the system. We develop a new method based on Seasonal-Trend decomposition with locally weighted regression (Loess) and likelihood ratio statistics to detect the changes. Experiments on daily rule-firing-count data collected from a real CDSS and known change-points show that our method improves the detection performance when compared with existing change-point detection methods.

Key-frame selection for automatic summarization of surveillance videos: a method of multiple change-point detection

Recent years have witnessed a drastic growth of various videos in real-life scenarios, and thus there is an increasing demand for a quick view of such videos in a constrained amount of time. In this paper, we focus on automatic summarization of surveillance videos and present a new key-frame selection method for this task. We first introduce a dissimilarity measure based on f-divergence by a symmetric strategy for multiple change-point detection and then use it to segment a given video sequence into a set of non-overlapping clips. Key frames are extracted from the resulting video clips by a typical clustering procedure for final video summary. Through experiments on a wide range of testing data, excellent performances, outperforming given state-of-the-art competitors, have been demonstrated which suggests good potentials of the proposed method in real-world applications.

Seismic status in Bangladesh

Seismic status in Bangladesh

Calculating shock arrival in expansion tubes and shock tunnels using Bayesian changepoint analysis

To understand the flow conditions generated in expansion tubes and shock tunnels, shock speeds are generally calculated based on shock arrival times at high-frequency wall-mounted pressure transducers. These calculations require that the shock arrival times are obtained accurately. This can be non-trivial for expansion tubes especially because pressure rises may be small and shock speeds high. Inaccurate shock arrival times can be a significant source of uncertainty. To help address this problem, this paper investigates two separate but complimentary techniques. Principally, it proposes using a Bayesian changepoint detection method to automatically calculate shock arrival, potentially reducing error and simplifying the shock arrival finding process. To compliment this, a technique for filtering the raw data without losing the shock arrival time is also presented and investigated. To test the validity of the proposed techniques, tests are performed using both a theoretical step change with different levels of noise and real experimental data. It was found that with conditions added to ensure that a real shock arrival time was found, the Bayesian changepoint analysis method was able to automatically find the shock arrival time, even for noisy signals.

Comparison of classic and novel change point detection methods for time series with changes in variance

Segmentation or change point detection is a very common topic in time series analysis, anomaly detection and pattern recognition. In our previous paper the time series generated by sensors with 3D accelerometers were analysed. It was noticed that such series consist of segments of independent and correlated observations. Hence the appropriate methods for change point detection for both data types must be implemented simultaneously.This paper provides an auxiliary comparison analysis which we intend to implement later for the above mentioned acceleration data.The available methods require usually a long execution time, so that it is time-consuming if several methods should be compared. In the framework of the present publication we want to give additional help for detecting a suitable change point detection method and for finding a good parameter setting. Our analysis is performed on simulated time series, that are normally distributed with constant but unknown mean and changes in variance.

Capturing correlation changes by applying kernel change point detection on the running correlations

Change point detection methods signal the occurrence of abrupt changes in a time series. Non-parametric approaches, such as the Gaussian kernel based change point (KCP) detection (Arlot et al., 2012), are especially attractive because they impose less assumptions on the data. Yet, a drawback of these methods is that most of them are sensitive to changes in the mean, the variance, etc., making them less sensitive to specific kinds of changes. We show that KCP can be adapted to detect a particular type of change only. We focus here on correlation change, which has been put forward by different theories but proved hard to trace in multivariate time series. We propose KCP-corr, which boils down to applying KCP on the running correlations. To confirm that KCP-corr is more sensitive than merely applying KCP on the raw data (KCP-raw), a simulation study was conducted in which the number of (noise) variables and the size of the correlation change were varied. KCP-corr emerged as the better method especially in the more difficult but realistic settings where the correlation change is minimal and/or noise variables are present. KCP-corr also outperforms Cusum, a non-parametric method that specifically targets the detection of correlation changes.

A Double Recursion for Calculating Moments of the Truncated Normal Distribution and its Connection to Change Detection

The integral ${\int }_{0}^{\infty }x^{m} e^{-\frac {1}{2}(x-a)^{2}}dx$ appears in likelihood ratios used to detect a change in the parameters of a normal distribution. As part of the mth moment of a truncated normal distribution, this integral is known to satisfy a recursion relation, which has been used to calculate the first four moments of a truncated normal. Use of higher order moments was rare. In more recent times, this integral has found important applications in methods of changepoint detection, with m going up to the thousands. The standard recursion formula entails numbers whose values grow quickly with m, rendering a low cap on computational feasibility. We present various aspects of dealing with the computational issues: asymptotics, recursion and approximation. We provide an example in a changepoint detection setting.

A Doubly Stochastic Change Point Detection Algorithm for Noisy Biological Signals.

Experimentally and clinically collected time series data are often contaminated with significant confounding noise, creating short, noisy time series. This noise, due to natural variability and measurement error, poses a challenge to conventional change point detection methods. We propose a novel and robust statistical method for change point detection for noisy biological time sequences. Our method is a significant improvement over traditional change point detection methods, which only examine a potential anomaly at a single time point. In contrast, our method considers all suspected anomaly points and considers the joint probability distribution of the number of change points and the elapsed time between two consecutive anomalies. We validate our method with three simulated time series, a widely accepted benchmark data set, two geological time series, a data set of ECG recordings, and a physiological data set of heart rate variability measurements of fetal sheep model of human labor, comparing it to three existing methods. Our method demonstrates significantly improved performance over the existing point-wise detection methods.

Automatic identification of near-stationary traffic states based on the PELT changepoint detection

The existence of stationary states during peak periods has been an underlying assumption of many studies on analysis, operations, control, and management of transportation networks. In Cassidy (1998), a method was proposed to manually identify near-stationary states by visually inspecting transformed curves of cumulative total vehicle counts and occupancies. Such near-stationary states are important for calibrating fundamental diagrams, identifying active bottlenecks and incidents, and quantifying capacity drop magnitudes. To the best of our knowledge, however, there lacks an automatic method that can be applied to efficiently identify near-stationary states from a large amount of data.In this study, we attempt to fill this gap. We start with definitions of steady, stationary, and equilibrium states and discuss their relations. Then we present a novel four-step method for automatically identifying near-stationary states from raw loop-detector data: first, the raw data are pre-processed to obtain healthy datasets, fill in missing values, and normalize averaged vehicle counts and occupancies to the same scale; second, daily time series are partitioned into multiple candidate intervals based on the pruned exact linear time (PELT) changepoint detection method; third, the characteristics of the candidate intervals are calculated; and finally, near-stationary states are selected based on modified Cassidy’s gap and duration criteria. We further close the loop by presenting an algorithm to automatically determine the penalty threshold in the second step and the gap threshold in fourth step to ensure the quantity as well as the quality of identified near-stationary states. In a case study, we apply the proposed method to identify near-stationary states from a large set of 30-s raw loop-detector data at a freeway mainline station. We verify the validity of identified near-stationary states both directly and indirectly. The results show that the identified near-stationary states are valid with high quality and the calibrated triangular fundamental diagram is well-fitted and physically meaningful. We finally conclude by discussing some future improvements and potential applications.

Bazı Değişim Noktası Algılama Yöntemlerinin İncelenmesi

Bazı Değişim Noktası Algılama Yöntemlerinin İncelenmesi

Identification of Temporal Transition of Functional States Using Recurrent Neural Networks from Functional MRI.

Dynamic functional connectivity analysis provides valuable information for understanding brain functional activity underlying different cognitive processes. Besides sliding window based approaches, a variety of methods have been developed to automatically split the entire functional MRI scan into segments by detecting change points of functional signals to facilitate better characterization of temporally dynamic functional connectivity patterns. However, these methods are based on certain assumptions for the functional signals, such as Gaussian distribution, which are not necessarily suitable for the fMRI data. In this study, we develop a deep learning based framework for adaptively detecting temporally dynamic functional state transitions in a data-driven way without any explicit modeling assumptions, by leveraging recent advances in recurrent neural networks (RNNs) for sequence modeling. Particularly, we solve this problem in an anomaly detection framework with an assumption that the functional profile of one single time point could be reliably predicted based on its preceding profiles within a stable functional state, while large prediction errors would occur around change points of functional states. We evaluate the proposed method using both task and resting-state fMRI data obtained from the human connectome project and experimental results have demonstrated that the proposed change point detection method could effectively identify change points between different task events and split the resting-state fMRI into segments with distinct functional connectivity patterns.

ASSESSING LANDSAT FRACTIONAL GROUND-COVER TIME SERIES ACROSS AUSTRALIA’S ARID RANGELANDS: SEPARATING GRAZING IMPACTS FROM CLIMATE VARIABILITY

Abstract. Suitable measures of grazing impacts on ground cover, that enable separation of the effects of climatic variations, are needed to inform land managers and policy makers across the arid rangelands of Australia. This work developed and tested a time-series, changepoint detection method for application to time series of vegetation fractional cover derived from Landsat data to identify irregular and episodic ground-cover growth cycles. Utilising the High Performance Computing power of the Google Cloud Compute Engine these cycles were segmented to distinguish grazing impacts from that of climate variability. A measure of grazing impact was developed using a multivariate technique to quantify the rate and degree of ground cover change. The method was successful in detecting both long term and short term growth cycles. Growth cycle detection was assessed against rainfall surplus measures indicating a relationship with high rainfall periods. During periods of ground cover decline, grazing utilisation was observed across four major grasslands. Ground cover change associated with grazing impacts was also assessed against field measurements of ground cover indicating a relationship between both field and remotely sensed ground cover. Cause and effects between grazing practices and ground cover resilience can now be explored in isolation to climatic drivers. This is important to the long term balance between ground cover utilisation and overall landscape function and resilience.

Change-Point Detection for Monitoring Clinical Decision Support Systems with a Multi-Process Dynamic Linear Model.

A clinical decision support system and its components may malfunction due to different reasons. The objective of this work is to develop computational methods that can help us to monitor the system and assure its proper operation by promptly detecting and analyzing changes in its behavior. We develop a new change-point detection method using the Multi-Process Dynamic Linear Model. The experiments on real and simulated data show that our method outperforms existing change-point detection methods, leading to higher accuracy and shorter delay in the detection.

Robust change point detection method via adaptive LAD-LASSO

Change point problem is one of the hot issues in statistics, econometrics, signal processing and so on. LAD estimator is more robust than OLS estimator, especially when datasets subject to heavy tailed errors or outliers. LASSO is a popular choice for shrinkage estimation. In the paper, we combine the two classical ideas together to put forward a robust detection method via adaptive LAD-LASSO to estimate change points in the mean-shift model. The basic idea is converting the change point estimation problem into variable selection problem with penalty. An enhanced two-step procedure is proposed. Simulation and a real example show that the novel method is really feasible and the fast and effective computation algorithm is easier to realize.

Parallelization of Change Point Detection.

The change point detection method ( Watkins , L. P. ; Yang , H. J. Phys. Chem. B 2005 , 109 , 617 ) allows the objective identification and isolation of abrupt changes along a data series. Because this method is grounded in statistical tests, it is particularly powerful for probing complex and noisy signals without artificially imposing a kinetics model. The original algorithm, however, has a time complexity of [Formula: see text], where N is the size of the data and is, therefore, limited in its scalability. This paper puts forth a parallelization of change point detection to address these time and memory constraints. This parallelization method was evaluated by applying it to changes in the mean of Gaussian-distributed data and found that time decreases superlinearly with respect to the number of processes (i.e., parallelization with two processes takes less than half of the time of one process). Moreover, there was minimal reduction in detection power. These results suggest that our parallelization algorithm is a viable scheme that can be implemented for other change point detection methods.

A travel time forecasting model based on change-point detection method

Travel time parameters obtained from road traffic sensors data play an important role in traffic management practice. A travel time forecasting model is proposed for urban road traffic sensors data based on the method of change-point detection in this paper. The first-order differential operation is used for preprocessing over the actual loop data; a change-point detection algorithm is designed to classify the sequence of large number of travel time data items into several patterns; then a travel time forecasting model is established based on autoregressive integrated moving average (ARIMA) model. By computer simulation, different control parameters are chosen for adaptive change point search for travel time series, which is divided into several sections of similar state.Then linear weight function is used to fit travel time sequence and to forecast travel time. The results show that the model has high accuracy in travel time forecasting.

Change-Point Detection Method for Clinical Decision Support System Rule Monitoring.

A clinical decision support system (CDSS) and its components can malfunction due to various reasons. Monitoring the system and detecting its malfunctions can help one to avoid any potential mistakes and associated costs. In this paper, we investigate the problem of detecting changes in the CDSS operation, in particular its monitoring and alerting subsystem, by monitoring its rule firing counts. The detection should be performed online, that is whenever a new datum arrives, we want to have a score indicating how likely there is a change in the system. We develop a new method based on Seasonal-Trend decomposition and likelihood ratio statistics to detect the changes. Experiments on real and simulated data show that our method has a lower delay in detection compared with existing change-point detection methods.