Using penalized contrasts for the change-point problem

By Marc Lavielle. Signal Processing, 2005

659 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/b070acbe937f1b0d8003d21f5deea20941bec76e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

Citations in the corpus

Inference for single and multiple change‐points in time series (2013, methodology): “On the basis of some numerical computations of Birgé and Massart (2001), Lavielle (2005) suggested that the constant c in (51) be c=2.5.”
Homogeneity and change-point detection tests for multivariate data using rank statistics (2011, background): “…L⋆ ≥ 1 change-points, if IL(n) is plotted against L, the resulting graph can be decomposed into two distinct regions: the first one, for L = 0, . . . , L⋆ where the criterion is growing rapidly; and the second one, for L = L⋆, . . . , Lmax, where the criterion is barely increasing (Lavielle, 2005).”
A robust approach for estimating change-points in the mean of an AR(1) process (2014, background): “Many model selection criteria have also been proposed to estimate the number of changes, mostly in the independent case (see, for example, [20,22,37,39]).”
A Modified Bayes Information Criterion with Applications to the Analysis of Comparative Genomic Hybridization Data (2007, methodology): “In comparison to the classical BIC and penalized likelihood methods (e.g., Tibshirani, 1996; Birgé and Massart, 2001; Gu and Wang, 2003; Lavielle, 2005), the modified BIC differs in the penalty term.”
A review of change point detection methods (2018, methodology): “We refer the reader to [89, 90] for theoretical results in this context.”
Selective review of offline change point detection methods (2020, background): “Historically, Opt 653 was introduced for a non-related problem [109] and later applied to change point 654 detection, in many different contexts, such as EEG recordings [65, 110], DNA 655 sequences [100, 111], tree growth monitoring [20], financial time-series [49, 76], 656 radar waveforms [112], et”
Wild binary segmentation for multiple change-point detection (2014, background): “…and locations of change-points for dependent ε t ’s, including the cases of strong mixing and long-range dependence; see also Lavielle (1999) for a discussion and some extensions of this result and Lavielle (2005) for some practical proposals regarding the adaptive choice of the penalty parameter.”
Multiple Change-Point Estimation With a Total Variation Penalty (2010, methodology): “…selecting the number of change points is usually performed thanks to a Schwarz-like penalty λ n K , where λ n is often calibrated on data (Lavielle and Moulines 2000; Lavielle 2005), or a penalty K ( a + b log ( n / K )) as in Lebarbier (2005), Massart (2005), where a and b are data-driven as well.”
changepoint: An R Package for Changepoint Analysis (2014, background): “The choice of appropriate penalty is still an open question and typically depends on many factors including the size of the changes and the length of segments, both of which are unknown prior to analysis (see Guyon and Yao (1999); Lavielle (2005); Birge and Massart (2007)).”
Animal movements in heterogeneous landscapes: identifying profitable places and homogeneous movement bouts. (2008, methodology): “Lavielle (2005) showed that using AIC or BIC strongly overestimates the number of breakpoints, thus creating the rationale for the use of another criterion.”
A SIMPLE BAYESIAN APPROACH TO MULTIPLE CHANGE-POINTS (2011, methodology): “Although this is the case when there are many possible but no clear-cut segments, there are sometimes subject-matter reasons for segmentation, especially if the signal actually consists of a few long segments as prescribed by conditions (C1) and (C2); see Siegmund (2004), Olshen et al. (2004) and La”
Multi-level functional responses for wildlife conservation: the case of threatened caribou in managed boreal forests (2012, methodology): “An algorithm was used to split the year into five distinct periods (Lavielle 2005).”