Explaining Time Series via Contrastive and Locally Sparse Perturbations

2024/01/16 by Zichuan Liu, Liu, Zichuan, Yingying Zhang +19 · 6 citations
Computer Science · Engineering · #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning in Healthcare #Time Series Analysis and Forecasting #Traffic Prediction and Management Techniques

paper · pdf · doi:10.48550/arxiv.2401.08552

openalex publication_date 2024/01/16 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Explaining multivariate time series is a compound challenge, as it requires identifying important locations in the time series and matching complex temporal patterns. Although previous saliency-based methods addressed the challenges, their perturbation may not alleviate the distribution shift issue, which is inevitable especially in heterogeneous samples. We present ContraLSP, a locally sparse model that introduces counterfactual samples to build uninformative perturbations but keeps distribution using contrastive learning. Furthermore, we incorporate sample-specific sparse gates to generate more binary-skewed and smooth masks, which easily integrate temporal trends and select the salient features parsimoniously. Empirical studies on both synthetic and real-world datasets show that ContraLSP outperforms state-of-the-art models, demonstrating a substantial improvement in explanation quality for time series data. The source code is available at \urlhttps://github.com/zichuan-liu/ContraLSP.

Cited by

• Delta-XAI: A Unified Framework for Explaining Prediction Changes in Online Time Series Monitoring
• TriShGAN: Enhancing Sparsity and Robustness in Multivariate Time Series Counterfactuals Explanation
• ShapeX: Shapelet-Driven Post Hoc Explanations for Time Series Classification Models
• Information Bottleneck Learning for Faithful Time Series Forecasting Explanations
• Time-RA: Towards Time Series Reasoning for Anomaly with LLM Feedback
• DeltaSHAP: Explaining Prediction Evolutions in Online Patient Monitoring with Shapley Values

Related

• Frequency-Masked Embedding Inference: A Non-Contrastive Approach for Time Series Representation Learning
• Learning Perturbations to Explain Time Series Predictions
• FreqRISE: Explaining time series using frequency masking
• Generating Sparse Counterfactual Explanations For Multivariate Time Series
• CLeaRForecast: Contrastive Learning of High-Purity Representations for Time Series Forecasting