Interpretable time series classification using linear models and multi-resolution multi-domain symbolic representations

  title={Interpretable time series classification using linear models and multi-resolution multi-domain symbolic representations},
  author={Thach Le Nguyen and Severin Gsponer and Iulia Ilie and Martin O'Reilly and Georgiana Ifrim},
  journal={Data Mining and Knowledge Discovery},
The time series classification literature has expanded rapidly over the last decade, with many new classification approaches published each year. Prior research has mostly focused on improving the accuracy and efficiency of classifiers, with interpretability being somewhat neglected. This aspect of classifiers has become critical for many application domains and the introduction of the EU GDPR legislation in 2018 is likely to further emphasize the importance of interpretable learning algorithms… 

The great multivariate time series classification bake off: a review and experimental evaluation of recent algorithmic advances

It is demonstrated that four classifiers are significantly more accurate than the benchmark dynamic time warping algorithm and that one of these recently proposed classifiers achieves significant improvement on the archive datasets in at least an order of magnitude less time than the other three.

Explainable AI for Time Series Classification: A Review, Taxonomy and Research Directions

This work presents the first extensive literature review on Explainable AI (XAI) for time series classification, and categorizes the research field through a taxonomy subdividing the methods into time points-based, subsequences-based and instance-based.

An Examination of the State-of-the-Art for Multivariate Time Series Classification

This work investigates the state-of-the-art for multivariate time series classification using the UEA MTSC benchmark, and proposes a simple statistics-based time series classifier that is as accurate as DTW, but is an order of magnitude faster.

Fast, Accurate and Interpretable Time Series Classification Through Randomization

Experiments show that r-STSF achieves state-of-the-art accuracy while being orders of magnitude faster than most existing TSC methods, and highlights that the best TSC method when classifying complex time series datasets.

A Reinforcement Learning-Informed Pattern Mining Framework for Multivariate Time Series Classification

This work proposes a reinforcement learning (RL) informed PAttern Mining framework (RLPAM) to identify interpretable yet important patterns for MTS classification and shows how RL informed patterns can be interpretable and can improve the understanding of septic shock progression.

MrSQM: Fast Time Series Classification with Symbolic Representations

MrSQM, a new time series classifier that uses multiple symbolic representations and efficient sequence mining, to extract important time series features, and shows that a fast symbolic transform combined with a simple feature selection strategy can be highly effective as compared to more sophisticated and expensive feature selection methods.

Instance-Based Counterfactual Explanations for Time Series Classification

The Native-Guide method retrieves and uses native in-samplecounterfactuals that already exist in the training data as "guides" for perturbation in time series counterfactual generation, coupled with both Euclidean and Dynamic Time Warping distance measures.

HIVE-COTE 2.0: a new meta ensemble for time series classification

This work proposes comprehensive changes to the HIVE-COTE algorithm which significantly improve its accuracy and usability, and introduces two novel classifiers, the Temporal Dictionary Ensemble and Diverse Representation Canonical Interval Forest, which replace existing ensemble members.

Discriminative Dictionary Learning for Time Series Classification

A novel and fast variable-length word generation method, incorporating the skip-bigram model for the construction of symbiotic word pairs, is proposed for the first time and a discriminative word filter with a dynamic threshold is designed to build the discrim inative word dictionary.



Time-Series Classification with COTE: The Collective of Transformation-Based Ensembles

Through extensive experimentation on 72 datasets, it is demonstrated that the simple collective formed by including all classifiers in one ensemble is significantly more accurate than any of its components and any other previously published TSC algorithm.

Deep learning for time series classification

The main objective of this thesis was to study and develop deep neural networks specifically constructed for the classification of time series data, and to carry out the first large scale experimental study allowing to compare the existing deep methods and to position them compared other non-deep learning based state-of-the-art methods.

Scalable time series classification

The Bag-Of-SFA-Symbols in Vector Space classifier is presented, which is significantly more accurate than 1-NN DTW while being multiple orders of magnitude faster and relevant for use cases like long or large amounts of time series or real-time analytics.

Time Series Classification by Sequence Learning in All-Subsequence Space

This work studies an efficient linear classification approach, SEQL, originally designed for classification of symbolic sequences, and proposes two adaptations of SEQL that can deal with X-axis offsets by learning variable-length symbolic words, and SAX-VFSEQL, which are linear classifiers in rich feature spaces.

The great time series classification bake off: a review and experimental evaluation of recent algorithmic advances

This work implemented 18 recently proposed algorithms in a common Java framework and compared them against two standard benchmark classifiers (and each other) by performing 100 resampling experiments on each of the 85 datasets, indicating that only nine of these algorithms are significantly more accurate than both benchmarks.

Deep learning for time series classification: a review

This article proposes the most exhaustive study of DNNs for TSC by training 8730 deep learning models on 97 time series datasets and provides an open source deep learning framework to the TSC community.

HIVE-COTE: The Hierarchical Vote Collective of Transformation-Based Ensembles for Time Series Classification

A common CNN structure is implemented and it is found that Flat-COTE is significantly more accurate than both deep learning approaches on 85 datasets, and represents a new state of the art for TSC.

The BOSS is concerned with time series classification in the presence of noise

This model combines the extraction of substructures with the tolerance to extraneous and erroneous data using a noise reducing representation of the time series to improve classification accuracies in diverse application areas and on the official UCR classification benchmark datasets by a large margin.

A Bag-of-Features Framework to Classify Time Series

A framework to classify time series based on a bag-of-features representation (TSBF) that provides a feature-based approach that can handle warping (although differently from DTW), and experimental results show that TSBF provides better results than competitive methods on benchmark datasets from the UCR time series database.

Multivariate Time Series Classification with WEASEL+MUSE

The novelty of WEASEL+MUSE lies in its specific way of extracting and filtering multivariate features from MTS by encoding context information into each feature, resulting in a small, yet very discriminative feature set useful for MTS classification.