Variation pattern classification of functional data

@article{Jiao2020VariationPC,
  title={Variation pattern classification of functional data},
  author={Shuhao Jiao and Ron D. Frostig and Hernando C. Ombao},
  journal={Canadian Journal of Statistics},
  year={2020}
}
A new classification method for functional data is developed for the case where different groups or classes of functions have similar mean functions but potentially different second moments. The proposed method, second moment-based functional classifier (SMFC), uses the Hilbert-Schmidt norm to measure the discrepancy between the second moment structure of the different groups. The proposed SMFC method is demonstrated to be sensitive to the discrepancy in the second moment structure and thus… 
View PDF on arXiv
2 Citations
Background Citations
1
Methods Citations
1

Figures and Tables from this paper

2 Citations

Covariance-based soft clustering of functional data based on the Wasserstein-Procrustes metric

A soft clustering methodology based on the Wasserstein-Procrustes distance, where the in-between cluster variability is penalised by a term proportional to the entropy of the partition matrix, which allows for clusters to overlap.

Break Point Detection for Functional Covariance

This work proposes a detection and testing procedure to find the change-points in functional covariance and establishes the asymptotic properties of the estimated change-point.

References

SHOWING 1-10 OF 43 REFERENCES

RGB-D object classification using covariance descriptors

This work introduces a new covariance based feature descriptor to be used on “colored” point clouds gathered by a mobile robot equipped with an RGB-D camera, and presents the notion of a covariance onRGB-D data.

Classification of multivariate non-stationary signals: The SLEX-shrinkage approach

Consistent Classification of Nonstationary Time Series Using Stochastic Wavelet Representations

This work proposes a method for classification using a bias-corrected nondecimated wavelet transform, which is illustrated using seismic signals and demonstrated to work very well in simulation studies.

Break Point Detection for Functional Covariance

This work proposes a detection and testing procedure to find the change-points in functional covariance and establishes the asymptotic properties of the estimated change-point.

Modeling Spectral Properties in Stationary Processes of Varying Dimensions with Applications to Brain Local Field Potential Signals

A technique to automatically identify frequency bands that carry significant spectral power in various frequency bands is developed and indicates that spectral information in the Beta frequency band demonstrated the greatest change in structure and complexity due to the stroke.

On local feature extraction for signal classification

Since the basis functions well-localized in the time-frequency plane are used as feature extractors, interpretation of the classification results becomes easier and more intuitive than using the conventional methods directly on the original coordinate system.

Achieving near perfect classification for functional data

It is shown that, in functional data classification problems, perfect asymptotic classification is often possible, making use of the intrinsic very high dimensional nature of functional data, which points to a marked contrast between classification for functional data and its counterpart in conventional multivariate analysis.

Correlation-Based Functional Clustering via Subspace Projection

A correlation-based functional clustering method is proposed for grouping curves with similar shapes. A correlation between two random functions defined through the functional inner product is used…

Bayesian Curve Classification Using Wavelets

An unified hierarchical model to encompass both the adaptive wavelet-based function estimation model and the logistic classification model is developed, couple together these two models are to borrow strengths from each other in a unified hierarchical framework.

Functional clustering and identifying substructures of longitudinal data

It is shown that, under the identifiability conditions derived, the "k"-centres FC method proposed can greatly improve cluster quality as compared with conventional clustering algorithms.