HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation
Journal articles

[Re] How Attention Can Create Synaptic Tags for the Learning of Working Memories in Sequential Tasks

Erwan Le Masson 1 Frédéric Alexandre 1
1 Mnemosyne - Mnemonic Synergy
LaBRI - Laboratoire Bordelais de Recherche en Informatique, Inria Bordeaux - Sud-Ouest, IMN - Institut des Maladies Neurodégénératives [Bordeaux]
Abstract : The reference paper introduces a new reinforcement learning model called Attention- Gated MEmory Tagging (AuGMEnT). The results presented suggest new approaches in understanding the acquisition of tasks requiring working memory and attentional feedback, as well as biologically plausible learning mechanisms. The model also improves on previous reinforcement learning schemes by allowing tasks to be expressed more naturally as a sequence of inputs and outputs. A Python implementation of the model is available on the author’s GitHub page which helped to verify the correctness of the computations. The script written for this replication also uses Python along with NumPy.
Document type :
Journal articles
Complete list of metadata

Contributor : Frédéric Alexandre Connect in order to contact the contributor
Submitted on : Friday, December 16, 2016 - 9:32:03 PM
Last modification on : Monday, December 20, 2021 - 4:50:15 PM
Long-term archiving on: : Tuesday, March 21, 2017 - 9:08:00 AM


Publisher files allowed on an open archive




Erwan Le Masson, Frédéric Alexandre. [Re] How Attention Can Create Synaptic Tags for the Learning of Working Memories in Sequential Tasks. The ReScience journal, GitHub, 2016, 2 (1), ⟨10.1371/journal.pcbi.1004060⟩. ⟨hal-01418735⟩



Record views


Files downloads