Evaluating parallel minibatch training for machine learning applications

Publikation: Beitrag in Buch/Bericht/TagungsbandKonferenzbeitragBegutachtung


The amount of data available for analytics applications continues to rise. At the same time, there are some application areas where security and privacy concerns prevent liberal dissemination of data. Both of these factors motivate the hypothesis that machine learning algorithms may benefit from parallelizing the training process (for large amounts of data) and/or distributing the training process (for sensitive data that cannot be shared). We investigate this hypothesis by considering two real-world machine learning tasks (logistic regression and sparse autoencoder), and empirically test how a model’s performance changes when its parameters are set to the arithmetic means of parameters of models trained on minibatches, i.e., horizontally split portions of the data set. We observe that iterating the minibatch training and parameter averaging process for a small number of times results in models with performance only slightly worse that of models trained on the full data sets.

TitelComputer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers
Redakteure/-innenRoberto Moreno-Diaz, Alexis Quesada-Arencibia, Franz Pichler
Herausgeber (Verlag)Springer
ISBN (Print)9783319747170
PublikationsstatusVeröffentlicht - 2018
Veranstaltung16th International Conference on Computer Aided Systems Theory, EUROCAST 2017 - Las Palmas de Gran Canaria, Spanien
Dauer: 19 Feb. 201724 Feb. 2017


NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Band10671 LNCS
ISSN (Print)0302-9743
ISSN (elektronisch)1611-3349


Konferenz16th International Conference on Computer Aided Systems Theory, EUROCAST 2017
OrtLas Palmas de Gran Canaria


Untersuchen Sie die Forschungsthemen von „Evaluating parallel minibatch training for machine learning applications“. Zusammen bilden sie einen einzigartigen Fingerprint.