A high-capacity model for one shot association learning in the brain
OPEN ACCESS
Author / Producer
Date
2014-11
Publication Type
Journal Article
ETH Bibliography
yes
Citations
Altmetric
OPEN ACCESS
Data
Rights / License
Abstract
We present a high-capacity model for one-shot association learning (hetero-associative memory) in sparse networks. We assume that basic patterns are pre-learned in networks and associations between two patterns are presented only once and have to be learned immediately. The model is a combination of an Amit-Fusi like network sparsely connected to a Willshaw type network. The learning procedure is palimpsest and comes from earlier work on one-shot pattern learning. However, in our setup we can enhance the capacity of the network by iterative retrieval. This yields a model for sparse brain-like networks in which populations of a few thousand neurons are capable of learning hundreds of associations even if they are presented only once. The analysis of the model is based on a novel result by Janson et al. on bootstrap percolation in random graphs.
Permanent link
Publication status
published
External links
Editor
Book title
Journal / series
Volume
8
Pages / Article No.
140
Publisher
Frontiers Media
Event
Edition / version
Methods
Software
Geographic location
Date collected
Date created
Subject
one shot learning; hetero-associative memory; relation learning; bootstrap percolation; iterative retrieval; stochastic Hebbian learning; memory capacity
Organisational unit
02803 - Collegium Helveticum / Collegium Helveticum
03325 - Folkers, Gerd (emeritus)
03672 - Steger, Angelika / Steger, Angelika