Selective Learning for Deep Time Series Forecasting

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Fu, Yisong, Shao, Zezhi, Yu, Chengqing, Li, Yujie, An, Zhulin, Wang, Qi, Xu, Yongjun, Wang, Fei
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917049259786240
author Fu, Yisong
Shao, Zezhi
Yu, Chengqing
Li, Yujie
An, Zhulin
Wang, Qi
Xu, Yongjun
Wang, Fei
author_facet Fu, Yisong
Shao, Zezhi
Yu, Chengqing
Li, Yujie
An, Zhulin
Wang, Qi
Xu, Yongjun
Wang, Fei
contents Benefiting from high capacity for capturing complex temporal patterns, deep learning (DL) has significantly advanced time series forecasting (TSF). However, deep models tend to suffer from severe overfitting due to the inherent vulnerability of time series to noise and anomalies. The prevailing DL paradigm uniformly optimizes all timesteps through the MSE loss and learns those uncertain and anomalous timesteps without difference, ultimately resulting in overfitting. To address this, we propose a novel selective learning strategy for deep TSF. Specifically, selective learning screens a subset of the whole timesteps to calculate the MSE loss in optimization, guiding the model to focus on generalizable timesteps while disregarding non-generalizable ones. Our framework introduces a dual-mask mechanism to target timesteps: (1) an uncertainty mask leveraging residual entropy to filter uncertain timesteps, and (2) an anomaly mask employing residual lower bound estimation to exclude anomalous timesteps. Extensive experiments across eight real-world datasets demonstrate that selective learning can significantly improve the predictive performance for typical state-of-the-art deep models, including 37.4% MSE reduction for Informer, 8.4% for TimesNet, and 6.5% for iTransformer.
format Preprint
id arxiv_https___arxiv_org_abs_2510_25207
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Selective Learning for Deep Time Series Forecasting
Fu, Yisong
Shao, Zezhi
Yu, Chengqing
Li, Yujie
An, Zhulin
Wang, Qi
Xu, Yongjun
Wang, Fei
Machine Learning
Benefiting from high capacity for capturing complex temporal patterns, deep learning (DL) has significantly advanced time series forecasting (TSF). However, deep models tend to suffer from severe overfitting due to the inherent vulnerability of time series to noise and anomalies. The prevailing DL paradigm uniformly optimizes all timesteps through the MSE loss and learns those uncertain and anomalous timesteps without difference, ultimately resulting in overfitting. To address this, we propose a novel selective learning strategy for deep TSF. Specifically, selective learning screens a subset of the whole timesteps to calculate the MSE loss in optimization, guiding the model to focus on generalizable timesteps while disregarding non-generalizable ones. Our framework introduces a dual-mask mechanism to target timesteps: (1) an uncertainty mask leveraging residual entropy to filter uncertain timesteps, and (2) an anomaly mask employing residual lower bound estimation to exclude anomalous timesteps. Extensive experiments across eight real-world datasets demonstrate that selective learning can significantly improve the predictive performance for typical state-of-the-art deep models, including 37.4% MSE reduction for Informer, 8.4% for TimesNet, and 6.5% for iTransformer.
title Selective Learning for Deep Time Series Forecasting
topic Machine Learning
url https://arxiv.org/abs/2510.25207