Dynamic Model Selection With its Applications to Novelty Detection

Yamanishi, Kenji; Maruyama, Y.

doi:10.1109/tit.2007.896890

Cited by 35 publications

(20 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Yamanishi and Maruyama [16,17] developed a theory of dynamic model selection (DMS) for tracking changes of statistical models from non-stationary data. Although this theory has been developed in a general setting for statistical model sequence selection, it can be applied to tracking of changes of clustering structures.…”

Section: Previous Workmentioning

confidence: 99%

“…1)An extension of DMS into a sequential clustering setting: In [16,17], the theory of DMS for estimating model sequences has been explored in the batch scenario where the whole data set is given at once, and the model sequence must be detected in a retrospective way. It has remained open how to extend the DMS algorithm into the sequential scenario where data are sequentially input and the model must be selected in a sequential fashion.…”

Section: Novelty and Significance Of This Papermentioning

confidence: 99%

See 1 more Smart Citation

Detecting changes of clustering structures using normalized maximum likelihood coding

Hirai

Yamanishi

2012

Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

Self Cite

View full text Add to dashboard Cite

We are concerned with the issue of detecting changes of clustering structures from multivariate time series. From the viewpoint of the minimum description length (MDL) principle, we propose an algorithm that tracks changes of clustering structures so that the sum of the code-length for data and that for clustering changes is minimum. Here we employ a Gaussian mixture model (GMM) as representation of clustering, and compute the code-length for data sequences using the normalized maximum likelihood (NML) coding. The proposed algorithm enables us to deal with clustering dynamics including merging, splitting, emergence, disappearance of clusters from a unifying view of the MDL principle. We empirically demonstrate using artificial data sets that our proposed method is able to detect cluster changes significantly more accurately than an existing statistical-test based method and AIC/BIC-based methods. We further use real customers' transaction data sets to demonstrate the validity of our algorithm in market analysis. We show that it is able to detect changes of customer groups, which correspond to changes of real market environments.

show abstract

Section: Previous Workmentioning

confidence: 99%

Section: Novelty and Significance Of This Papermentioning

confidence: 99%

Detecting changes of clustering structures using normalized maximum likelihood coding

Hirai

Yamanishi

2012

Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

Self Cite

View full text Add to dashboard Cite

show abstract

“…Fig.2 lists a number of studies for parameter estimation and model selection of the number of component algorithms [1,4,5,12,18,21] and their online/nonstationary learning extensions [16,15,22,27,28], including OMHMC.…”

Section: Online Model Selection Algorithmmentioning

confidence: 99%

Online heterogeneous mixture modeling with marginal and copula selection

Fujimaki¹,

Sogawa

Morinaga

2011

Proceedings of the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

View full text Add to dashboard Cite

This paper proposes an online mixture modeling methodology in which individual components can have different marginal distributions and dependency structures. Mixture models have been widely studied and applied to various application areas, including density estimation, fraud/failure detection, image segmentation, etc. Previous research has been almost exclusively focused on mixture models having components of a single type (e.g., a Gaussian mixture model.) However, recent growing needs for complicated data modeling necessitate the use of more flexible mixture models (e.g., a mixture of a lognormal distribution for medical costs and a Gaussian distribution for blood pressure, for medical analytics.) Our key ideas include: 1) separating marginal distributions and their dependencies using copulas and 2) online extension of a recently-developed "expectation minimization of description length," which enable us to efficiently learn types of both marginal distributions and copulas as well as their parameters. The proposed method provides not only good performance in applications, but also scalable, automatic model selection, which greatly reduces the intensive modeling costs in data mining processes. We show that the proposed method outperforms state-of-the-art methods in application to density estimation and to anomaly detection.

show abstract

“…The analysis of structural breaks is the main subject of [22],wherethegoalistodetectmalicious use of computer resources by looking for changes in the structure of consecutive best fitting models over a nontstationary time series. [22] refers to its approach as Dynamic Model Selection and uses an MDL coding scheme to find the best break point locations.…”

Section: Introductionmentioning

confidence: 99%

“…[22] refers to its approach as Dynamic Model Selection and uses an MDL coding scheme to find the best break point locations.…”

Section: Introductionmentioning

confidence: 99%

Scalable structural break detection

íLtetö¹,

Hansen²,

Germain-Renaud³

et al. 2012

Applied Soft Computing

View full text Add to dashboard Cite

This paper deals with a statistical model fitting procedure for non-stationary time series. This procedure selects the parameters of a piecewise autoregressive model using the Minimum Description Length principle. The existing chromosome representation of the piecewise autoregressive model and its corresponding optimisation algorithm are improved.F i r s t , w e s h o w t h a t o u r p r o p o s e d chromosome representation better captures the intrinsic properties of the piecewise autoregressive model. Second, we apply an optimisation algorithm, the Covariance Matrix Adaptation -Evolution Strategy, with which our setup converges faster to the optimal fit. Our proposed method achieves at least one order of magnitude performance improvement compared to the existing solution.

show abstract

Dynamic Model Selection With its Applications to Novelty Detection

Cited by 35 publications

References 19 publications

Detecting changes of clustering structures using normalized maximum likelihood coding

Detecting changes of clustering structures using normalized maximum likelihood coding

Online heterogeneous mixture modeling with marginal and copula selection

Scalable structural break detection

Contact Info

Product

Resources

About