MIX: AMulti-view Time-Frequency Interactive Explanation Framework for Time Series Classification

Deep learning models for time series classification (TSC) have achieved impressive performance, but explaining their decisions remains a significant challenge.Existing post-hoc explanation methods typically operate solely in the time domainand from a single-view perspective, limiting both faithfulness and robustness. Inthis work, we propose MIX (Multi-view Time-Frequency Interactive EXplanationFramework), a novel framework that helps to explain deep learning models ina multi-view setting by leveraging multi-resolution, time-frequency views constructed using the Haar Discrete Wavelet Transform (DWT). MIX introduces aninteractive cross-view refinement scheme, where explanation’s information fromone view is propagated across views to enhance overall interpretability. To alignwith user-preferred perspectives, we propose a greedy selection strategy that traverses the multi-view space to identify the most informative features. Additionally,we present OSIGV, a user-aligned segment-level attribution mechanism based onoverlapping windows for each view, and introduce keystone-first IG, a method thatrefines explanations in each view using additional information from another view.Extensive experiments across multiple TSC benchmarks and model architecturesdemonstrate that MIX significantly outperforms state-of-the-art (SOTA) methodsin terms of explanation faithfulness and robustness.