Patch-wise Structural Loss for Time Series Forecasting

Dilfira Kudrat, Zongxia Xie, Yanru Sun, Tianyu Jia, Qinghua Hu · 2025

Read Patch-wise Structural Loss for Time Series Forecasting by Dilfira Kudrat, Zongxia Xie, Yanru Sun, Tianyu Jia, Qinghua Hu (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-2503-00877.txt.

Description

Dilfira Kudrat and co-authors wrote a 2025 paper on a new loss function for forecasting time series. Standard losses grade each time step alone, blind to shape. Their patch wise loss grades whole chunks at once, comparing local mean, spread, and correlation between forecast and truth.

It pairs with the classic point loss rather than replacing it. Across many real data sets, top models trained with it forecast clearly better. The authors pitch shape aware grading as a general upgrade: a forecast can hug every point yet miss the pattern, and only a structural check catches that failure.

Catalog record

CREATORDilfira Kudrat, Zongxia Xie, Yanru Sun, Tianyu Jia, Qinghua Hu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE56 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-00877

Browse the full library · About the Tower of Babel