Efficient Time Series Forecasting via Hyper-Complex Models and Frequency Aggregation

Eyal Yakir, Dor Tsur, Haim Permuter · 2025

Read Efficient Time Series Forecasting via Hyper-Complex Models and Frequency Aggregation by Eyal Yakir, Dor Tsur, Haim Permuter (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-19983.txt.

Description

Eyal Yakir, Dor Tsur, and Haim Permuter present FIA-Net, a 2025 forecasting model built to read long time series in the frequency domain. Long-range links in a sequence are the classic hard part of forecasting, and chopping the series into windows for frequency analysis is a known help.

Their design uses a complex-valued network that blends neighboring windows where their frequency content meets. A further step treats the whole window set as hyper-complex numbers, an algebra that lets the model mix all windows at once with up to three times fewer parameters. On standard benchmarks, the method beat leading rivals on both accuracy and cost.

Catalog record

CREATOREyal Yakir, Dor Tsur, Haim Permuter
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-19983

Browse the full library · About the Tower of Babel