Hebbian errors in learning: An analysis using the Oja model

Rǎdulescu, Anca; Cox, Kingsley J. A.; Adams, Patrick W.

doi:10.1016/j.jtbi.2009.01.036

Cited by 18 publications

(35 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the absence of noise in the input patterns, introducing NSSP resulted in a decreased pattern recognition performance, similar to the loss of performance reported for unsupervised learning based on Oja’s rule [4]. However, in the presence of spatial noise in the input patterns, NSSP improved the pattern recognition performance [5].…”

supporting

confidence: 55%

Non-specific LTD at parallel fibre - Purkinje cell synapses in cerebellar cortex provides robustness against local spatial noise during pattern recognition

et al. 2011

View full text Add to dashboard Cite

show abstract

supporting

confidence: 55%

Non-specific LTD at parallel fibre - Purkinje cell synapses in cerebellar cortex provides robustness against local spatial noise during pattern recognition

et al. 2011

View full text Add to dashboard Cite

show abstract

“…In previous work we studied the effect of learning inaccuracy, or crosstalk, in simplified neural network models using a linear Hebbian learning rule, which is only sensitive to pairwise correlations (Cox and Adams, 2000; Adams and Cox, 2002a,b; Radulescu et al, 2009). We found that useful learning is always still possible provided that Hebbian adjustments retain some connection specificity, though it is degraded.…”

Section: Introductionmentioning

confidence: 99%

Hebbian crosstalk prevents nonlinear unsupervised learning

Cox

Adams²

2009

Front. Comput. Neurosci.

View full text Add to dashboard Cite

Learning is thought to occur by localized, activity-induced changes in the strength of synaptic connections between neurons. Recent work has shown that induction of change at one connection can affect changes at others (“crosstalk”). We studied the role of such crosstalk in nonlinear Hebbian learning using a neural network implementation of independent components analysis. We find that there is a sudden qualitative change in the performance of the network at a threshold crosstalk level, and discuss the implications of this for nonlinear learning from higher-order correlations in the neocortex.

show abstract

“…In apparent contrast to this common assumption, it has been shown that in brain areas such as the cerebellum synaptic plasticity can spread to neighboring inactive synapses2627282930313233. The presence of this kind of non-specific synaptic plasticity is expected to be detrimental for the recall of stored patterns34. We have investigated the storage and recall of input patterns in the presence and absence of non-specific long-term depression (nsLTD) in cerebellar PC models with different levels of complexity and biological realism.…”

Section: Discussionmentioning

confidence: 94%

“…This NO diffuses to neighboring synapses and compromises the synapse specificity of LTD2627282930313233. A recent theoretical study predicted that such non-specific plasticity would be detrimental for memory34.…”

mentioning

confidence: 99%

Nonspecific synaptic plasticity improves the recognition of sparse patterns degraded by local noise

Safaryan

Maex

Davey

et al. 2017

Sci Rep

View full text Add to dashboard Cite

Many forms of synaptic plasticity require the local production of volatile or rapidly diffusing substances such as nitric oxide. The nonspecific plasticity these neuromodulators may induce at neighboring non-active synapses is thought to be detrimental for the specificity of memory storage. We show here that memory retrieval may benefit from this non-specific plasticity when the applied sparse binary input patterns are degraded by local noise. Simulations of a biophysically realistic model of a cerebellar Purkinje cell in a pattern recognition task show that, in the absence of noise, leakage of plasticity to adjacent synapses degrades the recognition of sparse static patterns. However, above a local noise level of 20%, the model with nonspecific plasticity outperforms the standard, specific model. The gain in performance is greatest when the spatial distribution of noise in the input matches the range of diffusion-induced plasticity. Hence non-specific plasticity may offer a benefit in noisy environments or when the pressure to generalize is strong.

show abstract

Hebbian errors in learning: An analysis using the Oja model

Cited by 18 publications

References 51 publications

Non-specific LTD at parallel fibre - Purkinje cell synapses in cerebellar cortex provides robustness against local spatial noise during pattern recognition

Non-specific LTD at parallel fibre - Purkinje cell synapses in cerebellar cortex provides robustness against local spatial noise during pattern recognition

Hebbian crosstalk prevents nonlinear unsupervised learning

Nonspecific synaptic plasticity improves the recognition of sparse patterns degraded by local noise

Contact Info

Product

Resources

About