A Three-Threshold Learning Rule Approaches the Maximal Capacity of Recurrent Neural Networks.
Date
2015-08
Editors
Journal Title
Journal ISSN
Volume Title
Repository Usage Stats
views
downloads
Citation Stats
Abstract
Understanding the theoretical foundations of how memories are encoded and retrieved in neural populations is a central challenge in neuroscience. A popular theoretical scenario for modeling memory function is the attractor neural network scenario, whose prototype is the Hopfield model. The model simplicity and the locality of the synaptic update rules come at the cost of a poor storage capacity, compared with the capacity achieved with perceptron learning algorithms. Here, by transforming the perceptron learning rule, we present an online learning rule for a recurrent neural network that achieves near-maximal storage capacity without an explicit supervisory error signal, relying only upon locally accessible information. The fully-connected network consists of excitatory binary neurons with plastic recurrent connections and non-plastic inhibitory feedback stabilizing the network dynamics; the memory patterns to be memorized are presented online as strong afferent currents, producing a bimodal distribution for the neuron synaptic inputs. Synapses corresponding to active inputs are modified as a function of the value of the local fields with respect to three thresholds. Above the highest threshold, and below the lowest threshold, no plasticity occurs. In between these two thresholds, potentiation/depression occurs when the local field is above/below an intermediate threshold. We simulated and analyzed a network of binary neurons implementing this rule and measured its storage capacity for different sizes of the basins of attraction. The storage capacity obtained through numerical simulations is shown to be close to the value predicted by analytical calculations. We also measured the dependence of capacity on the strength of external inputs. Finally, we quantified the statistics of the resulting synaptic connectivity matrix, and found that both the fraction of zero weight synapses and the degree of symmetry of the weight matrix increase with the number of stored patterns.
Type
Department
Description
Provenance
Citation
Permalink
Published Version (Please cite this version)
Publication Info
Alemi, Alireza, Carlo Baldassi, Nicolas Brunel and Riccardo Zecchina (2015). A Three-Threshold Learning Rule Approaches the Maximal Capacity of Recurrent Neural Networks. PLoS Comput Biol, 11(8). p. e1004439. 10.1371/journal.pcbi.1004439 Retrieved from https://hdl.handle.net/10161/15109.
This is constructed from limited available data and may be imprecise. To cite this article, please review & use the official citation provided by the journal.
Collections
Scholars@Duke
![Brunel](https://scholars.duke.edu/profile-images/thumbnail200/0785756.jpg)
Nicolas Brunel
We use theoretical models of brain systems to investigate how they process and learn information from their inputs. Our current work focuses on the mechanisms of learning and memory, from the synapse to the network level, in collaboration with various experimental groups. Using methods from
statistical physics, we have shown recently that the synaptic
connectivity of a network that maximizes storage capacity reproduces
two key experimentally observed features: low connection probability
and strong overrepresentation of bidirectionnally connected pairs of
neurons. We have also inferred `synaptic plasticity rules' (a
mathematical description of how synaptic strength depends on the
activity of pre and post-synaptic neurons) from data, and shown that
networks endowed with a plasticity rule inferred from data have a
storage capacity that is close to the optimal bound.
Unless otherwise indicated, scholarly articles published by Duke faculty members are made available here with a CC-BY-NC (Creative Commons Attribution Non-Commercial) license, as enabled by the Duke Open Access Policy. If you wish to use the materials in ways not already permitted under CC-BY-NC, please consult the copyright owner. Other materials are made available here through the author’s grant of a non-exclusive license to make their work openly accessible.