Guardado en:
Detalles Bibliográficos
Autores principales: You, Jiang, Cela, Arben, Natowicz, René, Ouanounou, Jacob, Siarry, Patrick
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2406.13871
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917699547824128
author You, Jiang
Cela, Arben
Natowicz, René
Ouanounou, Jacob
Siarry, Patrick
author_facet You, Jiang
Cela, Arben
Natowicz, René
Ouanounou, Jacob
Siarry, Patrick
contents Forecasting multivariate time series is a computationally intensive task challenged by extreme or redundant samples. Recent resampling methods aim to increase training efficiency by reweighting samples based on their running losses. However, these methods do not solve the problems caused by heavy-tailed distribution losses, such as overfitting to outliers. To tackle these issues, we introduce a novel approach: a Gaussian loss-weighted sampler that multiplies their running losses with a Gaussian distribution weight. It reduces the probability of selecting samples with very low or very high losses while favoring those close to average losses. As it creates a weighted loss distribution that is not heavy-tailed theoretically, there are several advantages to highlight compared to existing methods: 1) it relieves the inefficiency in learning redundant easy samples and overfitting to outliers, 2) It improves training efficiency by preferentially learning samples close to the average loss. Application on real-world time series forecasting datasets demonstrate improvements in prediction quality for 1%-4% using mean square error measurements in channel-independent settings. The code will be available online after 1 the review.
format Preprint
id arxiv_https___arxiv_org_abs_2406_13871
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Robust Time Series Forecasting with Non-Heavy-Tailed Gaussian Loss-Weighted Sampler
You, Jiang
Cela, Arben
Natowicz, René
Ouanounou, Jacob
Siarry, Patrick
Machine Learning
Forecasting multivariate time series is a computationally intensive task challenged by extreme or redundant samples. Recent resampling methods aim to increase training efficiency by reweighting samples based on their running losses. However, these methods do not solve the problems caused by heavy-tailed distribution losses, such as overfitting to outliers. To tackle these issues, we introduce a novel approach: a Gaussian loss-weighted sampler that multiplies their running losses with a Gaussian distribution weight. It reduces the probability of selecting samples with very low or very high losses while favoring those close to average losses. As it creates a weighted loss distribution that is not heavy-tailed theoretically, there are several advantages to highlight compared to existing methods: 1) it relieves the inefficiency in learning redundant easy samples and overfitting to outliers, 2) It improves training efficiency by preferentially learning samples close to the average loss. Application on real-world time series forecasting datasets demonstrate improvements in prediction quality for 1%-4% using mean square error measurements in channel-independent settings. The code will be available online after 1 the review.
title Robust Time Series Forecasting with Non-Heavy-Tailed Gaussian Loss-Weighted Sampler
topic Machine Learning
url https://arxiv.org/abs/2406.13871