Seq2Tens: an efficient representation of sequences by low-rank tensor projections

Conference item : Poster

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 free algebra — to capture this non-commutativity. To address the innate computational complexity of this algebra, we use compositions of low-rank tensor projections. This yields modular and scalable building blocks that give state-of-the-art performance on standard benchmarks such as multivariate time series classification, mortality prediction and generative models for video.

Authors: Toth, C; Bonnier, P; Oberhauser, H

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: OpenReview
• Article number: 3363
• Publication date: 2020-09-28
• Acceptance date: 2021-01-12
• Event title: Ninth International Conference on Learning Representations (ICLR 2021)
• Event location: Virtual event
• Event website: https://iclr.cc/Conferences/2021
• Event start date: 2021-05-03
• Event end date: 2021-05-07
• Language: English
• Keywords: FFR
• Subtype: Poster