Abstract:Multivariate time series (MTS) analysis is an increasingly popular research topic in recent years due to the vast amount of MTS data that are being generated in numerous fields such as genomics research, health informatics, finance and abnormal detection. The particularity of the data makes it a challenging task, e.g., missing data, different sampling frequencies, and random noise. Moreover, each instance depends not only on its past values but also has some dependency on other instances, and there exist discriminatory order-dependent characteristics. To address these challenges, in this paper, we introduce an order-preserving metric learning framework for multivariate time series prediction. Specifically, we adopt quadruplet-wise constraints which can encompass pair-wise and triplet-wise constraints to model similarity from complex label relations. To preserve the inherent temporal relationships of the instances in MTS, order-preserving Wasserstein distance is integrated to the framework to measure dissimilarity between MTS data, where the inverse difference moment regularization enforces flow-network with local homogeneous structures and the KL-divergence with a prior distribution regularization prevents flow-network between instances with faraway temporal locations. Besides the regularizations on flow-network, the ground measurement of the Wasserstein distance is replaced by Mahalanobis distance to increase its discrimination capability. An alternating iteration strategy is proposed to jointly optimize the Mahalanobis distance matrix in the ground measurement and the flow-network of Wasserstein distance. Extensive experiments on real-world clinical data from critical care are provided to demonstrate the effectiveness of the proposed method on sepsis prediction task.

Discriminative Transformation for Multi-Dimensional Temporal Sequences.

Multilevel Dynamic Time Warping: A Parameter-Light Method for Fast Time Series Classification

Discriminative Dimensionality Reduction for Multi-Dimensional Sequences.

Transforming Multidimensional Time Series into Interpretable Event Sequences for Advanced Data Mining

Integrating Data-Driven Segmentation, Local Feature Extraction and Fisher Kernel Encoding to Improve Time Series Classification

Multivariate time series classification based on μσ-DWC feature and tree-structured M-SVM

Linear and Deep Order-Preserving Wasserstein Discriminant Analysis

Order-Preserving Wasserstein Discriminant Analysis

Linear Sequence Discriminant Analysis: A Model-Based Dimensionality Reduction Method for Vector Sequences

Order-Preserving Metric Learning for Mining Multivariate Time Series

Max-Min Distance Analysis by Using Sequential SDP Relaxation for Dimension Reduction

A Deep Multi-Task Representation Learning Method for Time Series Classification and Retrieval.

Sequence-Aware Factorization Machines for Temporal Predictive Analytics

Multiscale spatial‐temporal transformer with consistency representation learning for multivariate time series classification

Extreme Learning Machine for Time Sequence Classification

Dynamic Multi-Network Mining of Tensor Time Series

Heteroscedastic Max–Min Distance Analysis for Dimensionality Reduction

Contextual Dependency Vision Transformer for Spectrogram-Based Multivariate Time Series Analysis

metricDTW: local distance metric learning in Dynamic Time Warping

Heteroscedastic Max-Min Distance Analysis.

Mgformer: Multi-group transformer for multivariate time series classification