Saved in:
Bibliographic Details
Main Authors: Haeberle, Matthieu, van Gerwen, Puck, Laplaza, Ruben, Briling, Ksenia R., Weinreich, Jan, Eisenbrand, Friedrich, Corminboeuf, Clemence
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2410.16122
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913814381854720
author Haeberle, Matthieu
van Gerwen, Puck
Laplaza, Ruben
Briling, Ksenia R.
Weinreich, Jan
Eisenbrand, Friedrich
Corminboeuf, Clemence
author_facet Haeberle, Matthieu
van Gerwen, Puck
Laplaza, Ruben
Briling, Ksenia R.
Weinreich, Jan
Eisenbrand, Friedrich
Corminboeuf, Clemence
contents Integer linear programming (ILP) is an elegant approach to solve linear optimization problems, naturally described using integer decision variables. Within the context of physics-inspired machine learning applied to chemistry, we demonstrate the relevance of an ILP formulation to select molecular training sets for predictions of size-extensive properties. We show that our algorithm outperforms existing unsupervised training set selection approaches, especially when predicting properties of molecules larger than those present in the training set. We argue that the reason for the improved performance is due to the selection that is based on the notion of local similarity (i.e., per-atom) and a unique ILP approach that finds optimal solutions efficiently. Altogether, this work provides a practical algorithm to improve the performance of physics-inspired machine learning models and offers insights into the conceptual differences with existing training set selection approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2410_16122
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Integer linear programming for unsupervised training set selection in molecular machine learning
Haeberle, Matthieu
van Gerwen, Puck
Laplaza, Ruben
Briling, Ksenia R.
Weinreich, Jan
Eisenbrand, Friedrich
Corminboeuf, Clemence
Chemical Physics
Machine Learning
Integer linear programming (ILP) is an elegant approach to solve linear optimization problems, naturally described using integer decision variables. Within the context of physics-inspired machine learning applied to chemistry, we demonstrate the relevance of an ILP formulation to select molecular training sets for predictions of size-extensive properties. We show that our algorithm outperforms existing unsupervised training set selection approaches, especially when predicting properties of molecules larger than those present in the training set. We argue that the reason for the improved performance is due to the selection that is based on the notion of local similarity (i.e., per-atom) and a unique ILP approach that finds optimal solutions efficiently. Altogether, this work provides a practical algorithm to improve the performance of physics-inspired machine learning models and offers insights into the conceptual differences with existing training set selection approaches.
title Integer linear programming for unsupervised training set selection in molecular machine learning
topic Chemical Physics
Machine Learning
url https://arxiv.org/abs/2410.16122