Saved in:
Bibliographic Details
Main Authors: Subiñas, Sergio Muñiz, González, Manuel L., Gómez, Jorge Ruiz, Ali, Alejandro Mata, Martín, Jorge Martínez, Hernando, Miguel Franco, García-Vico, Ángel Miguel
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.16075
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914100816117760
author Subiñas, Sergio Muñiz
González, Manuel L.
Gómez, Jorge Ruiz
Ali, Alejandro Mata
Martín, Jorge Martínez
Hernando, Miguel Franco
García-Vico, Ángel Miguel
author_facet Subiñas, Sergio Muñiz
González, Manuel L.
Gómez, Jorge Ruiz
Ali, Alejandro Mata
Martín, Jorge Martínez
Hernando, Miguel Franco
García-Vico, Ángel Miguel
contents This work introduces a post-training quantization (PTQ) method for dense neural networks via a novel ADAROUND-based QUBO formulation. Using the Frobenius distance between the theoretical output and the dequantized output (before the activation function) as the objective, an explicit QUBO whose binary variables represent the rounding choice for each weight and bias is obtained. Additionally, by exploiting the structure of the coefficient QUBO matrix, the global problem can be exactly decomposed into $n$ independent subproblems of size $f+1$, which can be efficiently solved using some heuristics such as simulated annealing. The approach is evaluated on MNIST, Fashion-MNIST, EMNIST, and CIFAR-10 across integer precisions from int8 to int1 and compared with a round-to-nearest traditional quantization methodology.
format Preprint
id arxiv_https___arxiv_org_abs_2510_16075
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Optimization of the quantization of dense neural networks from an exact QUBO formulation
Subiñas, Sergio Muñiz
González, Manuel L.
Gómez, Jorge Ruiz
Ali, Alejandro Mata
Martín, Jorge Martínez
Hernando, Miguel Franco
García-Vico, Ángel Miguel
Machine Learning
Artificial Intelligence
This work introduces a post-training quantization (PTQ) method for dense neural networks via a novel ADAROUND-based QUBO formulation. Using the Frobenius distance between the theoretical output and the dequantized output (before the activation function) as the objective, an explicit QUBO whose binary variables represent the rounding choice for each weight and bias is obtained. Additionally, by exploiting the structure of the coefficient QUBO matrix, the global problem can be exactly decomposed into $n$ independent subproblems of size $f+1$, which can be efficiently solved using some heuristics such as simulated annealing. The approach is evaluated on MNIST, Fashion-MNIST, EMNIST, and CIFAR-10 across integer precisions from int8 to int1 and compared with a round-to-nearest traditional quantization methodology.
title Optimization of the quantization of dense neural networks from an exact QUBO formulation
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2510.16075