Dataset Information

Stationary log-normal distribution of weights stems from spontaneous ordering in adaptive node networks.

ABSTRACT: Experimental evidence recently indicated that neural networks can learn in a different manner than was previously assumed, using adaptive nodes instead of adaptive links. Consequently, links to a node undergo the same adaptation, resulting in cooperative nonlinear dynamics with oscillating effective link weights. Here we show that the biological reality of stationary log-normal distribution of effective link weights in neural networks is a result of such adaptive nodes, although each effective link weight varies significantly in time. The underlying mechanism is a stochastic restoring force emerging from a spontaneous temporal ordering of spike pairs, generated by strong effective link preceding by a weak one. In addition, for feedforward adaptive node networks the number of dynamical attractors can scale exponentially with the number of links. These results are expected to advance deep learning capabilities and to open horizons to an interplay between adaptive node rules and the distribution of network link weights.

SUBMITTER: Uzan H

PROVIDER: S-EPMC6117314 | biostudies-literature | 2018 Aug

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Stationary log-normal distribution of weights stems from spontaneous ordering in adaptive node networks.

Uzan Herut H Sardi Shira S Goldental Amir A Vardi Roni R Kanter Ido I

Scientific reports 20180830 1

Experimental evidence recently indicated that neural networks can learn in a different manner than was previously assumed, using adaptive nodes instead of adaptive links. Consequently, links to a node undergo the same adaptation, resulting in cooperative nonlinear dynamics with oscillating effective link weights. Here we show that the biological reality of stationary log-normal distribution of effective link weights in neural networks is a result of such adaptive nodes, although each effective l ...[more]

PMID: 30166579

Similar Datasets

Project description:Virtually all studies of phage infections investigate bacteria growing exponentially in rich media. In nature, however, phages largely encounter non-growing cells. Bacteria entering stationary phase often activate well-studied stress defense mechanisms that drastically alter the cell, facilitating its long-term survival. An understanding of phage-host interactions in such conditions is of major importance from both an ecological and therapeutic standpoint. Here, we show that bacteriophage T4 can efficiently bind to, infect and kill E. coli in stationary phase, both in the presence and absence of a functional stationary-phase sigma factor, and explore the response of T4-infected stationary phase cells to the addition of fresh nutrients 5 or 24 h after that infection. An unexpected new mode of response has been identified. "Hibernation" mode is a persistent but reversible dormant state in which the infected cells make at least some phage enzymes, but halt phage development until appropriate nutrients become available before producing phage particles. Our evidence indicates that the block in hibernation mode occurs after the middle-mode stage of phage development; host DNA breakdown and the incorporation of the released nucleotides into phage DNA indicate that the enzymes of the nucleotide synthesizing complex, under middle-mode control, have been made and assembled into a functional state. Once fresh glucose and amino acids become available, the standard lytic infection process rapidly resumes and concentrations of up to 10(11) progeny phage (an average of about 40 phage per initially present cell) are produced. All evidence is consistent with the hibernation-mode control point lying between middle mode and late mode T4 gene expression. We have also observed a "scavenger" response, where the infecting phage takes advantage of whatever few nutrients are available to produce small quantities of progeny within 2 to 5 h after infection. The scavenger response seems able to produce no more than an average of one phage per originally available cell, and few if any further progeny are produced by cells in this mode even if fresh nutrients are made available later.

Dataset Information

Stationary log-normal distribution of weights stems from spontaneous ordering in adaptive node networks.

Publications

Stationary log-normal distribution of weights stems from spontaneous ordering in adaptive node networks.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets