Training Binary Layers by Self-Shrinking of Sigmoid Slope: Application to Fast MRI Acquisition

F. Martinini, A. Enttsel, A. Marchioni, M. Mangia, F. Pareschi, R. Rovatti, G. Setti

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Deep Neural Networks (DNN) have become popular and widespread because they combine computational power and flexibility, but they may present critical hyper-parameters that need to be tuned before the model can be trained. Recently, the use of trainable binary masks in the field of Magnetic Resonance Imaging (MRI) acquisition brought new state-of-the-art results, but with the disadvantage of introducing a bulky hyper-parameter, which tuning is usually time-consuming. We present a novel callback-based method that is applied during training and turns the tuning problem into a triviality, also bringing non-negligible performance improvements. We test our method on the fastMRI dataset.
Original languageEnglish (US)
Title of host publicationBioCAS 2022 - IEEE Biomedical Circuits and Systems Conference: Intelligent Biomedical Systems for a Better Future, Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages665-669
Number of pages5
ISBN (Print)9781665469173
DOIs
StatePublished - Jan 1 2022
Externally publishedYes

Fingerprint

Dive into the research topics of 'Training Binary Layers by Self-Shrinking of Sigmoid Slope: Application to Fast MRI Acquisition'. Together they form a unique fingerprint.

Cite this