Abstract
Motivated by the potential for parallel implementation of batch-based algorithms and the accelerated convergence achievable with approximated second order information a limited memory version of the BFGS algorithm has been receiving increasing attention in recent years for large neural network training problems. As the shape of the cost function is generally not quadratic and only becomes approximately quadratic in the vicinity of a minimum, the use
of second order information by L-BFGS can be unreliable during the initial phase of training, i.e. when far from a minimum. Therefore, to control the in
uence of second order information as training progresses, we propose a multi-batch L-BFGS algorithm, namely MB-AM, that gradually increases its trust in the curvature information by implementing a progressive storage and use of curvature data through a development-based increase (dev-increase) scheme. Using six discriminative modelling benchmark problems we show empirically that MB-AM has slightly faster convergence and, on average, achieves better solutions than the standard multi-batch L-BFGS algorithm when training MLP and CNN models.
of second order information by L-BFGS can be unreliable during the initial phase of training, i.e. when far from a minimum. Therefore, to control the in
uence of second order information as training progresses, we propose a multi-batch L-BFGS algorithm, namely MB-AM, that gradually increases its trust in the curvature information by implementing a progressive storage and use of curvature data through a development-based increase (dev-increase) scheme. Using six discriminative modelling benchmark problems we show empirically that MB-AM has slightly faster convergence and, on average, achieves better solutions than the standard multi-batch L-BFGS algorithm when training MLP and CNN models.
Original language | English |
---|---|
Pages (from-to) | 8199-8204 |
Number of pages | 6 |
Journal | IFAC-PapersOnLine |
Volume | 53 |
Issue number | 2 |
DOIs | |
Publication status | Published - 14 Apr 2021 |
Event | 21st World Congress of the International Federation of Automatic Control 2020 - Berlin, Germany Duration: 12 Jul 2020 → 17 Jul 2020 https://www.ifac2020.org/ |
Bibliographical note
Will be available Diamond Open Access at https://www.journals.elsevier.com/ifac-papersonlineFingerprint
Dive into the research topics of 'An Adaptive Memory Multi-Batch L-BFGS Algorithm for Neural Network Training'. Together they form a unique fingerprint.Student theses
-
Advances in machine learning for sustainable manufacturing
Zocco, F. (Author), Liu, X. (Supervisor) & McLoone, S. (Supervisor), Dec 2021Student thesis: Doctoral Thesis › Doctor of Philosophy