Exploring the impact of post-training rounding in regression models

0585135 - ÚI 2025 RIV DE eng J - Journal Article
Kalina, Jan
Exploring the impact of post-training rounding in regression models.
Applications of Mathematics. Roč. 69, č. 2 (2024), s. 257-271. ISSN 0862-7940. E-ISSN 1572-9109
R&D Projects: GA ČR(CZ) GA22-02067S
Institutional support: RVO:67985807
Keywords : supervised learning * trained model * perturbations * effect of rounding * low-precision arithmetic
OECD category: Statistics and probability
Impact factor: 0.7, year: 2022
Method of publishing: Open access
https://doi.org/10.21136/AM.2024.0090-23

Post-training rounding, also known as quantization, of estimated parameters stands as a widely adopted technique for mitigating energy consumption and latency in machine learning models. This theoretical endeavor delves into the examination of the impact of rounding estimated parameters in key regression methods within the realms of statistics and machine learning. The proposed approach allows for the perturbation of parameters through an additive error with values within a specified interval. This method is elucidated through its application to linear regression and is subsequently extended to encompass radial basis function networks, multilayer perceptrons, regularization networks, and logistic regression, maintaining a consistent approach throughout.
Permanent Link: https://hdl.handle.net/11104/0352867