Papers › TSEM: Temporally Weighted Spatiotemporal Explainable Neural Network for Multivariate...

TSEM: Temporally Weighted Spatiotemporal Explainable Neural Network for Multivariate Time Series

25 May 2022arXiv:2205.13012archive 2025-07-28

Anh-Duy Pham, Anastassia Kuestenmacher, Paul G. Ploeger

Deep learning has become a one-size-fits-all solution for technical and business domains thanks to its flexibility and adaptability. It is implemented using opaque models, which unfortunately undermines the outcome trustworthiness. In order to have a better understanding of the behavior of a system, particularly one driven by time series, a look inside a deep learning model so-called posthoc eXplainable Artificial Intelligence (XAI) approaches, is important. There are two major types of XAI for time series data, namely model-agnostic and model-specific. Model-specific approach is considered in this work. While other approaches employ either Class Activation Mapping (CAM) or Attention Mechanism, we merge the two strategies into a single system, simply called the Temporally Weighted Spatiotemporal Explainable Neural Network for Multivariate Time Series (TSEM). TSEM combines the capabilities of RNN and CNN models in such a way that RNN hidden units are employed as attention weights for the CNN feature maps temporal axis. The result shows that TSEM outperforms XCM. It is similar to STAM in terms of accuracy, while also satisfying a number of interpretability criteria, including causality, fidelity, and spatiotemporality.

PaperPDFCode

Code

a11to1n3/tsem officialpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

ClassificationExplainable Artificial Intelligence (XAI)Explainable artificial intelligenceTime SeriesTime Series AnalysisTime Series Classification

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Time Series Classification ArticularyWordRecognition TSEM Accuracy 0.557 #1 of 1 Archive leaderboard report
Time Series Classification BasicMotions TSEM Accuracy 0.925 #1 of 1 Archive leaderboard report
Time Series Classification Cricket TSEM Accuracy 0.722 #1 of 1 Archive leaderboard report
Time Series Classification ERing TSEM Accuracy 0.844 #1 of 1 Archive leaderboard report
Time Series Classification EigenWorms TSEM % Test Accuracy 42 #7 of 8 Archive leaderboard report
Time Series Classification EthanolConcentration TSEM Accuracy 0.395 #1 of 1 Archive leaderboard report
Time Series Classification FaceDetection TSEM Accuracy 0.513 #3 of 3 Archive leaderboard report
Time Series Classification Handwriting TSEM Accuracy 0.117 #1 of 1 Archive leaderboard report
Time Series Classification Heartbeat TSEM Accuracy 0.746 #3 of 3 Archive leaderboard report
Time Series Classification Libras TSEM Accuracy 0.372 #10 of 10 Archive leaderboard report
Time Series Classification NATOPS TSEM Accuracy 0.833 #1 of 1 Archive leaderboard report
Time Series Classification RacketSports TSEM Accuracy 0.77 #1 of 1 Archive leaderboard report
Time Series Classification SelfRegulationSCP2 TSEM Accuracy 0.756 #1 of 1 Archive leaderboard report
Time Series Classification StandWalkJump TSEM Accuracy 0.467 #1 of 1 Archive leaderboard report
Time Series Classification UCI Epileptic Seizure Recognition TSEM Accuracy 0.891 #1 of 1 Archive leaderboard report
Time Series Classification UWave TSEM Accuracy 0.831 #9 of 10 Archive leaderboard report
Time Series Classification pendigits TSEM Accuracy 0.686 #4 of 4 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections