Learning Soft Sparse Shapes for Efficient Time-Series Classification

2025-05-11

AI Summary

Existing shapelet-based time series classification methods rely on hard-thresholded feature selection, which risks discarding informative subsequences and ignores the heterogeneous contributions of different shapelets to classification. This paper proposes a soft-sparse shapelet modeling framework that abandons binary selection, instead retaining and differentially weighting all candidate subsequences. Our approach features: (1) dual modules for soft shapelet sparsification and soft shapelet learning; (2) a learnable, class-specific router that governs an expert network for fine-grained, interpretable pattern modeling; and (3) end-to-end differentiable shapelet optimization integrating soft attention weighting, class-conditional routing, shared experts, and shapelet-to-sequence transformation. Evaluated on multiple multiclass benchmark datasets, our method achieves significant improvements over state-of-the-art approaches, simultaneously attaining higher classification accuracy and clear, shapelet-level interpretability.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Shapelets are discriminative subsequences (or shapes) with high interpretability in time series classification. Due to the time-intensive nature of shapelet discovery, existing shapelet-based methods mainly focus on selecting discriminative shapes while discarding others to achieve candidate subsequence sparsification. However, this approach may exclude beneficial shapes and overlook the varying contributions of shapelets to classification performance. To this end, we propose a extbf{Soft} sparse extbf{Shape}s ( extbf{SoftShape}) model for efficient time series classification. Our approach mainly introduces soft shape sparsification and soft shape learning blocks. The former transforms shapes into soft representations based on classification contribution scores, merging lower-scored ones into a single shape to retain and differentiate all subsequence information. The latter facilitates intra- and inter-shape temporal pattern learning, improving model efficiency by using sparsified soft shapes as inputs. Specifically, we employ a learnable router to activate a subset of class-specific expert networks for intra-shape pattern learning. Meanwhile, a shared expert network learns inter-shape patterns by converting sparsified shapes into sequences. Extensive experiments show that SoftShape outperforms state-of-the-art methods and produces interpretable results.

Problem

• Improves efficiency in time-series classification using soft shapes
• Addresses limitations of traditional shapelet-based sparsification methods
• Enhances interpretability and retains beneficial subsequence information

Innovation

• Soft shape sparsification retains all subsequence information
• Learnable router activates class-specific expert networks
• Shared expert network learns inter-shape patterns