Mixed-integer quadratic programming reformulations of multi-task learning models

In this manuscript, we consider well-known multi-task learning (MTL) models from the literature for linear regression problems, such as clustered MTL or weakly constrained MTL. We propose novel reformulations of the training problem for these models, based on mixed-integer quadratic programming (MIQ...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Mathematics in engineering Ročník 5; číslo 1; s. 1 - 16
Hlavní autoři: Lapucci, Matteo, Pucci, Davide
Médium: Journal Article
Jazyk:angličtina
Vydáno: AIMS Press 01.03.2023
Témata:
clustered mtl
global optimization
miqp
multitask learning
weakly constrained mtl
ISSN:2640-3501, 2640-3501
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In this manuscript, we consider well-known multi-task learning (MTL) models from the literature for linear regression problems, such as clustered MTL or weakly constrained MTL. We propose novel reformulations of the training problem for these models, based on mixed-integer quadratic programming (MIQP) techniques. We show that our approach allows to drive the optimization process up to certified global optimality, exploiting popular off-the-shelf software solvers. By computational experiments on both synthetic and real-world datasets, we show that this strategy generally leads to improvements in terms of the predictive performance of the models, if compared to the classical local optimization techniques, based on alternating minimization strategies, that are usually employed. We also suggest a number of possible extensions of our model that should further improve the quality of the obtained regressors, introducing, for example, sparsity and features selection elements.
ISSN:2640-3501
2640-3501
DOI:10.3934/mine.2023020