Papers › Seq2Tens: An Efficient Representation of Sequences by Low-Rank Tensor Projections

Seq2Tens: An Efficient Representation of Sequences by Low-Rank Tensor Projections

12 Jun 2020ICLR 2021 1arXiv:2006.07027archive 2025-07-28

Csaba Toth, Patric Bonnier, Harald Oberhauser

Sequential data such as time series, video, or text can be challenging to analyse as the ordered structure gives rise to complex dependencies. At the heart of this is non-commutativity, in the sense that reordering the elements of a sequence can completely change its meaning. We use a classical mathematical object -- the tensor algebra -- to capture such dependencies. To address the innate computational complexity of high degree tensors, we use compositions of low-rank tensor projections. This yields modular and scalable building blocks for neural networks that give state-of-the-art performance on standard benchmarks such as multivariate time series classification and generative models for video.

PaperPDFConference PDFCode

Code

tgcsaba/seq2tens officialmentioned in papermentioned on GitHubtf 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

ImputationTime SeriesTime Series AnalysisTime Series Classificationtensor algebra

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Imputation HMNIST GP-VAE (B-NLST) AUROC 0.962 #1 of 1 Archive leaderboard report
Imputation HMNIST GP-VAE (B-NLST) MSE 0.092 #1 of 1 Archive leaderboard report
Imputation HMNIST GP-VAE (B-NLST) NLL 0.251 #1 of 1 Archive leaderboard report
Imputation PhysioNet Challenge 2012 GP-VAE (B-NLST) AUROC 0.743 #1 of 1 Archive leaderboard report
Imputation Sprites GP-VAE (B-NLST) MSE 0.002 #1 of 1 Archive leaderboard report
Time Series Classification AUSLAN FCN-SNLST Accuracy 0.993 #1 of 9 Archive leaderboard report
Time Series Classification AUSLAN SNLST Accuracy 0.969 #4 of 9 Archive leaderboard report
Time Series Classification ArabicDigits FCN-SNLST Accuracy 0.993 #3 of 10 Archive leaderboard report
Time Series Classification ArabicDigits SNLST Accuracy 0.968 #10 of 10 Archive leaderboard report
Time Series Classification CMUsubject16 SNLST Accuracy 1 #4 of 9 Archive leaderboard report
Time Series Classification CMUsubject16 FCN-SNLST Accuracy 1 #5 of 9 Archive leaderboard report
Time Series Classification CharacterTrajectories FCN-SNLST Accuracy 0.994 #2 of 9 Archive leaderboard report
Time Series Classification CharacterTrajectories SNLST Accuracy 0.957 #5 of 9 Archive leaderboard report
Time Series Classification DigitShapes SNLST Accuracy 1 #7 of 9 Archive leaderboard report
Time Series Classification DigitShapes FCN-SNLST Accuracy 1 #8 of 9 Archive leaderboard report
Time Series Classification ECG FCN-SNLST Accuracy 0.860 #2 of 9 Archive leaderboard report
Time Series Classification ECG SNLST Accuracy 0.842 #4 of 9 Archive leaderboard report
Time Series Classification JapaneseVowels FCN-SNLST Accuracy 0.980 #9 of 10 Archive leaderboard report
Time Series Classification JapaneseVowels SNLST Accuracy 0.979 #10 of 10 Archive leaderboard report
Time Series Classification KickvsPunch SNLST Accuracy 1 #2 of 9 Archive leaderboard report
Time Series Classification KickvsPunch FCN-SNLST Accuracy 1 #3 of 9 Archive leaderboard report
Time Series Classification Libras FCN-SNLST Accuracy 0.957 #2 of 10 Archive leaderboard report
Time Series Classification Libras SNLST Accuracy 0.773 #7 of 10 Archive leaderboard report
Time Series Classification NetFlow FCN-SNLST Accuracy 0.960 #1 of 9 Archive leaderboard report
Time Series Classification NetFlow SNLST Accuracy 0.793 #9 of 9 Archive leaderboard report
Time Series Classification PEMS FCN-SNLST Accuracy 0.857 #1 of 8 Archive leaderboard report
Time Series Classification PEMS SNLST Accuracy 0.747 #7 of 8 Archive leaderboard report
Time Series Classification PenDigits SNLST Accuracy 0.954 #2 of 8 Archive leaderboard report
Time Series Classification PenDigits FCN-SNLST Accuracy 0.953 #4 of 8 Archive leaderboard report
Time Series Classification SHAPES SNLST Accuracy 1 #7 of 9 Archive leaderboard report
Time Series Classification SHAPES FCN-SNLST Accuracy 1 #8 of 9 Archive leaderboard report
Time Series Classification UWave FCN-SNLST Accuracy 0.969 #3 of 10 Archive leaderboard report
Time Series Classification UWave SNLST Accuracy 0.938 #7 of 10 Archive leaderboard report
Time Series Classification Wafer FCN-SNLST Accuracy 0.989 #4 of 10 Archive leaderboard report
Time Series Classification Wafer SNLST Accuracy 0.981 #7 of 10 Archive leaderboard report
Time Series Classification WalkvsRun FCN-SNLST Accuracy 1 #8 of 9 Archive leaderboard report
Time Series Classification WalkvsRun SNLST Accuracy 1 #9 of 9 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.

Methods

Batch NormalizationConvolutionFCNGaussian ProcessMax Pooling

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