• Corpus ID: 7698983

Inductive Conformal Martingales for Change-Point Detection

  title={Inductive Conformal Martingales for Change-Point Detection},
  author={Denis Volkhonskiy and Evgeny Burnaev and Ilia Nouretdinov and Alexander Gammerman and Vladimir Vovk},
We consider the problem of quickest change-point detection in data streams. [] Key Method Instead we propose a new method for change-point detection based on Inductive Conformal Martingales, which requires only the independence and identical distribution of observations.

Using inductive conformal martingales for addressing concept drift in data stream classification

This paper investigates the use of Inductive Conformal Martingales (ICM) with the histogram betting function for detecting the occurrence of concept drift (CD) in data stream classification and proposes a new approach, which is much more computationally efficient than alternative ICM approaches.

A histogram based betting function for conformal martingales

This paper investigates the use of Conformal Martingales (CM) for providing a numerical indication of how likely it is that the exchangeability assumption holds on a set of data and compares its computational efficiency and its performance with a kernel betting function and the Kolmogorov-Smirnoff test.

Testing Exchangeability With Martingale for Change-Point Detection

An additive martingale is introduced, which is more amenable for designing exchangeability tests by exploiting the Hoeffding-Azuma lemma, and an online algorithm based on Beta distribution parametrization for constructing this betting function is discussed.

Conformal Kernel Expected Similarity for Anomaly Detection in Time-Series data

A modification of the EXPoSE algorithm for anomaly detection in time series data is proposed, which produces a probabilistic score of abnormality, which could allow an analyst to choose the anomaly threshold based on the desired false alarm rate.

Conformal prediction beyond exchangeability

These algorithms are provably robust, with substantially less loss of coverage when exchangeability is violated due to distribution drift or other challenging features of real data, while also achieving the same coverage guarantees as existing conformal prediction methods if the data points are in fact exchangeable.

Conformal prediction for time series

  • Chen XuYao Xie
  • Computer Science
  • 2020
A computationally efficient algorithm called EnbPI is introduced that wraps around ensemble predictors, which is closely related to conformal prediction (CP) but does not require data exchangeability.

Conformal prediction for dynamic time-series

A computationally efficient algorithm called EnbPI is introduced that wraps around ensemble predictors, which is closely related to conformal prediction (CP) but does not require data exchangeability.

Conformal prediction interval for dynamic time-series

A method to build distribution-free prediction intervals for time-series based on conformal inference that wraps around any ensemble estimator to construct sequential prediction intervals is developed, which is easy to implement, scalable to producing arbitrarily many prediction intervals sequentially, and well-suited to a wide range of regression functions.

Power and limitations of conformal martingales

This paper, accompanying my poster at ISIPTA 2019 (5 July 2019), poses the problem of investigating the power and limitations of conformal martingales as a means of detecting deviations from randomness and gives several preliminary results in the direction of efficiency.



Ensembles of detectors for online detection of transient changes

This work proposes a learning paradigm and specific implementations of ensembles for change detection of short-term (transient) changes in observed time series and demonstrates by means of numerical experiments that the performance of an ensemble is superior to that of the conventional change-point detection procedures.

A martingale framework for concept change detection in time-varying data streams

  • S. Ho
  • Computer Science
  • 2005
This paper extends the idea of testing exchangeability online to a martingale framework to detect concept changes in time-varying data streams and shows that bothMartingale tests are effective in detecting concept changesIn time-VaryingData streams simulated using two synthetic data sets and three benchmark data sets.

Nonparametric decomposition of quasi-periodic time series for change-point detection

A multicomponent time series model and an effective online decomposition algorithm to approximate the components of the models and propose two online change-point detection schemes corresponding to two real-world scenarios.

Average Run Lengths of an Optimal Method of Detecting a Change in Distribution.

It is of interest to declare that a change took place (to raise an alarm) as soon as possible after its occurrence, subject to a restriction on the rate of false detections.

Detection of abrupt changes: theory and application

A unified framework for the design and the performance analysis of the algorithms for solving change detection problems and links with the analytical redundancy approach to fault detection in linear systems are established.

Sequential Detection of Transient Changes

A suboptimal sequential transient change detection algorithm is proposed based on a window-limited cumulative sum (CUSUM) test and a lower and an upper bound for the false alarm rate are proposed.

Sequential Analysis: Hypothesis Testing and Changepoint Detection

This book covers the theoretical developments and applications of sequential hypothesis testing and sequential quickest changepoint detection in a wide range of engineering and environmental domains and explains how the theoretical aspects influence the hypothesisTesting and changepoint Detection problems as well as the design of algorithms.

Plug-in martingales for testing exchangeability on-line

This paper extends known techniques for constructing exchangeability martingales and shows that the new method is competitive with the martingale introduced before, and investigates the performance of the testing method on two benchmark datasets, USPS and Statlog Satellite data.

A generalized Shiryayev sequential probability ratio test for change detection and isolation

  • D. MalladiJ. Speyer
  • Mathematics
    Proceedings of 35th IEEE Conference on Decision and Control
  • 1996
It is shown that for a certain criterion of optimality, this generalized Shiryayev SPRT detects and isolates a change in hypothesis in the conditionally independent measurement sequence in minimum time, unlike the Wald SPRT, which assumes the entire measurement sequence to correspond to a single hypothesis.

Model selection for anomaly detection

This paper generalizes several kernel selection methods from binary-class case to the case of one-class classification and performs extensive comparison of these approaches using both synthetic and real-world data.