Localized Random Shapelets
Mael Guilleme,Simon Malinowski,Romain Tavenard,Xavier Renard +3 more
- 20 Sep 2019
- pp 85-97
TL;DR: This paper designs an interpretable shapelet model that takes into account the localization of the shapelets in the time series and designs a hierarchical feature selection process using regularization that has competitive performance compared to state-of-the-art shapelet-based classifiers, while providing better interpretability.
read more
Abstract: Shapelet models have attracted a lot of attention from researchers in the time series community, due in particular to its good classification performance. However, such models only inform about the presence/absence of local temporal patterns. Structural information about the localization of these patterns is ignored. In addition, end-to-end learning shapelet models tend to generate meaningless shapelets, leading to poorly interpretable models. In this paper, we aim at designing an interpretable shapelet model that takes into account the localization of the shapelets in the time series. Time series are transformed into feature vectors composed of both a distance and a localization information. Then, we design a hierarchical feature selection process using regularization. This process can be tuned to select, for each shapelet, either only its distance information or both distance and localization information. It is hence possible for every selected shapelet to analyze whether only the presence or the presence and the localization contributed to the decision process improving interpretability of the decision. Experiments show that this feature selection process has competitive performance compared to state-of-the-art shapelet-based classifiers, while providing better interpretability.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Figures
Fig. 2: Overview of our localized random shapelet model. Blue circles indicate distance features while orange ones correspond to location features. For each shapelet, a group is formed whose weights are denoted β(k) (where k is the shapelet index). Note that the number of hidden layers may vary from one application to the other. 
Fig. 6: The distribution of the localization (left) and distance (right) for the most important (first row) and second most important (second row) shapelet in TwoPatterns dataset. 
Fig. 5: Four most important shapelets (in red) extracted by our method from the TwoPatterns dataset. ![Fig. 1: Comparison between shapelets extracted by the Learning Time-Series Shapelets (LS) algorithm and our Localized Random Shapelets (LRS) approach. This Figure has been generated using tslearn implementation of LS [14].](/figures/fig-1-comparison-between-shapelets-extracted-by-the-learning-15hgxg0l.png)
Fig. 1: Comparison between shapelets extracted by the Learning Time-Series Shapelets (LS) algorithm and our Localized Random Shapelets (LRS) approach. This Figure has been generated using tslearn implementation of LS [14]. 
Fig. 8: Critical diagrams of the performance against the baselines. 
Fig. 7: Error rates comparison on 85 UCR datasets between LRS with ssgl regularization against LRS with lasso regularization.
![Fig. 1: Comparison between shapelets extracted by the Learning Time-Series Shapelets (LS) algorithm and our Localized Random Shapelets (LRS) approach. This Figure has been generated using tslearn implementation of LS [14].](/figures/fig-1-comparison-between-shapelets-extracted-by-the-learning-15hgxg0l.webp)
Citations
Explainable AI for Time Series Classification: A Review, Taxonomy and Research Directions
01 Jan 2022
TL;DR: In this article , the authors present an extensive literature review on explainable AI for time series classification, categorize the research field through a taxonomy subdividing the methods into time points-based, subsequences-based and instance-based.
GENDIS: Genetic Discovery of Shapelets.
TL;DR: In this article, a new paradigm for shapelet discovery is proposed, which is based on evolutionary computation, which can allow escaping from local optima more easily and supports non-differentiable objectives.
16
Random Dilated Shapelet Transform: A New Approach for Time Series Shapelets
TL;DR: In this article , a new formulation of time series shapelets including the notion of dilation is presented, and a new shapelet feature is introduced to enhance their discriminative power for classification.
GENDIS: GENetic DIscovery of Shapelets
TL;DR: A new paradigm for shapelet discovery is proposed, which is based on evolutionary computation and is gradient-free, which could allow escaping from local optima more easily and supports non-differentiable objectives.
9
Localized shapelets selection for interpretable time series classification
Jiahui Chen,Yuan Wan +1 more
TL;DR: This paper proposes a lo calized shapelets selection approach for interpretable time series classification using a location measure and distance measure to evaluate the discriminative ability of each shapelet candidate, and then the shapelet transformation process also integrates the location information of shapelets to provide a more interpretable insight in the classification result.
6
References
Regression Shrinkage and Selection via the Lasso
TL;DR: A new method for estimation in linear models called the lasso, which minimizes the residual sum of squares subject to the sum of the absolute value of the coefficients being less than a constant, is proposed.
•Proceedings Article
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Sergey Ioffe,Christian Szegedy +1 more
- 06 Jul 2015
TL;DR: Applied to a state-of-the-art image classification model, Batch Normalization achieves the same accuracy with 14 times fewer training steps, and beats the original model by a significant margin.
•Proceedings Article
Deep Sparse Rectifier Neural Networks
Xavier Glorot,Antoine Bordes,Yoshua Bengio +2 more
- 14 Jun 2011
TL;DR: This paper shows that rectifying neurons are an even better model of biological neurons and yield equal or better performance than hyperbolic tangent networks in spite of the hard non-linearity and non-dierentiabil ity.
A Sparse-Group Lasso
TL;DR: A regularized model for linear regression with ℓ1 andℓ2 penalties is introduced and it is shown that it has the desired effect of group-wise and within group sparsity.
1.5K
The great time series classification bake off: a review and experimental evaluation of recent algorithmic advances
TL;DR: 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.
1.5K