Explaining Time Series via Contrastive and Locally Sparse Perturbations
This work provides a method for explaining time series data, which is incremental as it builds on existing saliency-based approaches to handle distribution shifts.
The paper tackles the challenge of explaining multivariate time series by addressing distribution shift issues in heterogeneous samples, resulting in ContraLSP, which outperforms state-of-the-art models with substantial improvements in explanation quality.
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 \url{https://github.com/zichuan-liu/ContraLSP}.