BSAT: B-Spline Adaptive Tokenizer for Long-Term Time Series Forecasting

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Reinwardt, Maximilian, Eichelbeck, Michael, Althoff, Matthias
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909979986886656
author Reinwardt, Maximilian
Eichelbeck, Michael
Althoff, Matthias
author_facet Reinwardt, Maximilian
Eichelbeck, Michael
Althoff, Matthias
contents Long-term time series forecasting using transformers is hampered by the quadratic complexity of self-attention and the rigidity of uniform patching, which may be misaligned with the data's semantic structure. In this paper, we introduce the \textit{B-Spline Adaptive Tokenizer (BSAT)}, a novel, parameter-free method that adaptively segments a time series by fitting it with B-splines. BSAT algorithmically places tokens in high-curvature regions and represents each variable-length basis function as a fixed-size token, composed of its coefficient and position. Further, we propose a hybrid positional encoding that combines a additive learnable positional encoding with Rotary Positional Embedding featuring a layer-wise learnable base: L-RoPE. This allows each layer to attend to different temporal dependencies. Our experiments on several public benchmarks show that our model is competitive with strong performance at high compression rates. This makes it particularly well-suited for use cases with strong memory constraints.
format Preprint
id arxiv_https___arxiv_org_abs_2601_00698
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle BSAT: B-Spline Adaptive Tokenizer for Long-Term Time Series Forecasting
Reinwardt, Maximilian
Eichelbeck, Michael
Althoff, Matthias
Machine Learning
Long-term time series forecasting using transformers is hampered by the quadratic complexity of self-attention and the rigidity of uniform patching, which may be misaligned with the data's semantic structure. In this paper, we introduce the \textit{B-Spline Adaptive Tokenizer (BSAT)}, a novel, parameter-free method that adaptively segments a time series by fitting it with B-splines. BSAT algorithmically places tokens in high-curvature regions and represents each variable-length basis function as a fixed-size token, composed of its coefficient and position. Further, we propose a hybrid positional encoding that combines a additive learnable positional encoding with Rotary Positional Embedding featuring a layer-wise learnable base: L-RoPE. This allows each layer to attend to different temporal dependencies. Our experiments on several public benchmarks show that our model is competitive with strong performance at high compression rates. This makes it particularly well-suited for use cases with strong memory constraints.
title BSAT: B-Spline Adaptive Tokenizer for Long-Term Time Series Forecasting
topic Machine Learning
url https://arxiv.org/abs/2601.00698