Too much information is no information - how machine learning and feature selection could help in understanding the motor control of pointing

Aug 6, 2026

Upload by Logic

This study develops machine learning methods for multivariate analysis in motor control research, where large datasets are often analyzed univariately. High-dimensional electromyogram (EMG) vectors from arm and trunk muscles were recorded while subjects pointed at angles above and below the gravity-neutral horizontal plane. Linear Discriminant Analysis (LDA) performed binary classifications by pointing angle, yielding a composite index of muscular adjustments. A post-classification feature selection strategy compared class representations to identify feature combinations most altered by task conditions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages14

Reading Time35 min

AccessFree to download

Introduction

Machine learning for multivariate motor control analysis

Feature selection for identifying altered feature combinations

EMG data collection and classification approach

TYPE Original Research PUBLISHED 20 July 2023 DOI 10. 3389/fdata.2023.921355 OPEN ACCESS EDITED BY Guy Cheron, Université libre de Bruxelles, Belgium REVIEWED BY Hui Zhou, Nanjing University of Science and Technology, China Sabir Jacquir, Université Paris-Saclay, France *CORRESPONDENCE Elizabeth Thomas [Elizabeth.Thomas@u-bourgogne.fr](Elizabeth.Thomas@u-bourgogne.fr) RECEIVED 15 April 2022 ACCEPTED 16 June 2023 PUBLISHED 20 July 2023 CITATION Thomas E, Ali FB, Tolambiya A, Chambellant Fand Gaveau J (2023) Too much information is no information: how machine learning and feature selection could help in understanding the motor control of pointing. Front. Big Data 6:921355 . doi: 10.3389/fdata.2023.921355 COPYRIGHT © 2023 Thomas, Ali, Tolambiya, Chambellant and Gaveau. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Too much information is no information: how machine learning and feature selection could help in understanding the motor control of pointing Elizabeth Thomas1*, Ferid Ben Ali2 , Arvind Tolambiya3 , Florian Chambellant1 and Jérémie Gaveau1 1 INSERMU1093, UFR STAPS, Université de Bourgogne Franche Comté, Dijon, France, 2 School of Engineering and Computer Science, University of Hertfordshire, Hatﬁeld, United Kingdom, 3Applied Intelligence Hub, Accenture Solutions Private Ltd., Hyderabad, Telangana, India The aim of this study was to develop the use of Machine Learning techniques as a means of multivariate analysis in studies of motor control. These studies generate a huge amount of data, the analysis of which continues to be largely univariate. We propose the use of machine learning classiﬁcation and featureselection as a means of uncovering feature combinations that are altered between conditions. High dimensional electromyogram (EMG) vectors were generated as several arm and trunk muscles were recorded while subjects pointed at various angles above and below the gravity neutral horizontal plane. We used Linear Discriminant Analysis (LDA) to carry out binary classiﬁcations between the EMG vectors for pointing at a particular angle, vs. pointing at the gravity neutral direction. Classiﬁcation success provided a composite index of muscular adjustments for various task constraints—in this case, pointing angles. In order to ﬁnd the combination of features that were signiﬁcantly altered between task conditions, we conducted a post classiﬁcation feature selection i.e., investigated which combination of features had allowed for the classiﬁcation. Feature selection was done by comparing the representations of each category created by LDA for the classiﬁcation. In other words computing the di􀀀erence between the representations of each class. We propose that this approach will help with comparing high dimensional EMG patterns in two ways; (i) quantifying the e􀀀ectsof the entire pattern rather than using single arbitrarily deﬁned variables and (ii) identifying the parts of the patterns that convey the most information regarding the investigated e􀀀ects. KEYWORDS motor control, machine learning, feature selection, pointing, explainable machine learning 1. Introduction Movement takes place through the contractions of several muscles which then cause the displacement of several body segments. Much pain and energy therefore goes into the simultaneous collection of the variables connected with motor control studies. Curiously despite the energy invested in the synchronization of the data collection and the big data tables created by such experiments, the analysis of it largely take

FAQs

What problem does the study target in motor control research?

It targets the tendency to analyze motor control experiments largely in a univariate way despite the creation of large datasets.

How were the EMG data and conditions defined?

Subjects produced pointing movements at various angles above and below the gravity-neutral horizontal plane while multiple arm and trunk muscles were recorded as high-dimensional EMG vectors.

How does the method identify which features change across task conditions?

The approach uses LDA to classify EMG vectors, then performs post-classification feature selection by comparing representations across the categories to quantify differences and determine which feature combinations are altered.

Aug 6, 2026

Upload by Logic