Remember to Correct the Bias When Using Deep Learning for Regression!
Publikation: Bidrag til tidsskrift › Tidsskriftartikel › Forskning › fagfællebedømt
Standard
Remember to Correct the Bias When Using Deep Learning for Regression! / Igel, Christian; Oehmcke, Stefan.
I: KI - Kunstliche Intelligenz, Bind 37, Nr. 1, 2023, s. 33-40.Publikation: Bidrag til tidsskrift › Tidsskriftartikel › Forskning › fagfællebedømt
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - JOUR
T1 - Remember to Correct the Bias When Using Deep Learning for Regression!
AU - Igel, Christian
AU - Oehmcke, Stefan
N1 - Funding Information: The authors acknowledge support by the Villum Foundation through the project Deep Learning and Remote Sensing for Unlocking Global Ecosystem Resource Dynamics (DeReEco, project number 34306) and the Pioneer Centre for AI, DNRF grant number P1. Publisher Copyright: © 2023, The Author(s).
PY - 2023
Y1 - 2023
N2 - When training deep learning models for least-squares regression, we cannot expect that the training error residuals of the final model, selected after a fixed training time or based on performance on a hold-out data set, sum to zero. This can introduce a systematic error that accumulates if we are interested in the total aggregated performance over many data points (e.g., the sum of the residuals on previously unseen data). We suggest adjusting the bias of the machine learning model after training as a default post-processing step, which efficiently solves the problem. The severeness of the error accumulation and the effectiveness of the bias correction are demonstrated in exemplary experiments.
AB - When training deep learning models for least-squares regression, we cannot expect that the training error residuals of the final model, selected after a fixed training time or based on performance on a hold-out data set, sum to zero. This can introduce a systematic error that accumulates if we are interested in the total aggregated performance over many data points (e.g., the sum of the residuals on previously unseen data). We suggest adjusting the bias of the machine learning model after training as a default post-processing step, which efficiently solves the problem. The severeness of the error accumulation and the effectiveness of the bias correction are demonstrated in exemplary experiments.
KW - Bias correction
KW - Deep learning
KW - Regression
U2 - 10.1007/s13218-023-00801-0
DO - 10.1007/s13218-023-00801-0
M3 - Journal article
AN - SCOPUS:85152908339
VL - 37
SP - 33
EP - 40
JO - KI - Künstliche Intelligenz
JF - KI - Künstliche Intelligenz
SN - 0933-1875
IS - 1
ER -
ID: 347311460