Beyond Hebb: Exclusive-OR and Biological Learning

Klemm, Konstantin; Bornholdt, Stefan; Schuster, Hans‐Uwe

doi:10.1103/physrevlett.84.3013

Cited by 22 publications

(44 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Synaptic plasticity and neuronal plasticity have different effects on a network level [1,2]. Synaptic plasticity affects the transmission of a signal through a link, a synapse, connecting two nodes, neurons, in the network.…”

Section: Discussionmentioning

confidence: 99%

Error correction and fast detectors implemented by ultrafast neuronal plasticity

2014

View full text Add to dashboard Cite

We experimentally show that the neuron functions as a precise time-integrator, where the accumulated changes in neuronal response latencies, under complex and random stimulation patterns, are solely a function of a global quantity, the average time-lag between stimulations. In contrast, momentary leaps in the neuronal response latency follow trends of consecutive stimulations, indicating ultra-fast neuronal plasticity. On a circuit level, this ultra-fast neuronal plasticity phenomenon implements error-correction mechanisms and fast detectors for misplaced stimulations. Additionally, at moderate/high stimulation rates this phenomenon destabilizes/stabilizes a periodic neuronal activity disrupted by misplaced stimulations. I. INTRODUCTIONOn the network and circuit level, both synaptic and neuronal plasticity are present. These two distinct types of plasticity have different effects on the dynamics of a network [1,2]. On one hand, synaptic plasticity has been vastly researched, from the single neuron to the network level, specifically long and short-term plasticity. Shortterm synaptic plasticity reflects an increase (facilitation) and decrease (depression) in the probability of neurotransmitter release [3,4]. It affects the speed of synaptic signal transmission and can last from hundreds of milliseconds to seconds [3,5]. This phenomenon varies enormously depending on the neuronal and synaptic features as well as on the neuron's recent history of activity [6][7][8]. Neuronal plasticity, on the other hand, was examined mainly on the single neuron level [9,10], hence investigation of this phenomenon is still demanded.Short-term synaptic plasticity, in the form of facilitation and depression (FAD), is suggested to carry critical computational functions in neural circuits [1,11,12], thus one can hypothesize that neuronal plasticity carries similar computational functions. This hypothesis was not experimentally verified on the network level, and in addition its enormous variation seems to prevent reliable information processing [13,14]. Here we experimentally demonstrate neuronal ultra-fast plasticity on a time scale of several milliseconds and its applications to advanced computational tasks.

show abstract

Section: Discussionmentioning

confidence: 99%

Error correction and fast detectors implemented by ultrafast neuronal plasticity

2014

View full text Add to dashboard Cite

show abstract

“…al [10] (table I): 1) an exploratory version by setting β = 10, and 2) a greedy version by setting Fig. 9.…”

Section: Methodsmentioning

confidence: 99%

“…al. [10] used to solve the XOR problem will now be presented. In this work [10], the authors selected the node to activate stochastically where the probability of firing for each node j is:…”

Section: Minibrainmentioning

confidence: 99%

An Improved Minibrain That Learns Through Both Positive and Negative Feedback

Phua

Blair

2006

The 2006 IEEE International Joint Conference on Neural Network Proceedings

View full text Add to dashboard Cite

show abstract

“…Based on the experimental findings by Frey et al 7 and Otmakova et al 13 , Bak and Chialvo 14,15 as well as Klemm et al 16 suggested biologically inspired learning rules for neural networks that combine unsupervised Hebbian (homosynaptic) with reinforcement learning. We call this kind of combination of Hebbian and reinforcement learning Hebb-like learning to indicate that the learning rule is different from Hebb, but contains nevertheless characteristics which are biological plausible.…”

Section: Overview Of Biological and Artificial Learning In Neural Netmentioning

confidence: 99%

A Heterosynaptic Learning Rule for Neural Networks

Emmert‐Streib

2006

Int. J. Mod. Phys. C

View full text Add to dashboard Cite

In this article we intoduce a novel stochastic Hebb-like learning rule for neural networks that is neurobiologically motivated. This learning rule combines features of unsupervised (Hebbian) and supervised (reinforcement) learning and is stochastic with respect to the selection of the time points when a synapse is modified. Moreover, the learning rule does not only affect the synapse between pre-and postsynaptic neuron, which is called homosynaptic plasticity, but effects also further remote synapses of the pre-and postsynaptic neuron. This more complex form of synaptic plasticity has recently come under investigations in neurobiology and is called heterosynaptic plasticity. We demonstrate that this learning rule is useful in training neural networks by learning parity functions including the exclusive-or (XOR) mapping in a multilayer feed-forward network. We find, that our stochastic learning rule works well, even in the presence of noise. Importantly, the mean learning time increases with the number of patterns to be learned polynomially, indicating efficient learning.

show abstract

Beyond Hebb: Exclusive-OR and Biological Learning

Cited by 22 publications

References 20 publications

Error correction and fast detectors implemented by ultrafast neuronal plasticity

Error correction and fast detectors implemented by ultrafast neuronal plasticity

An Improved Minibrain That Learns Through Both Positive and Negative Feedback

A Heterosynaptic Learning Rule for Neural Networks

Contact Info

Product

Resources

About