On a 3-class MNIST subset, a competitive-Hebbian + weight-perturbation rule achieves higher estimated mutual information per nonsilent synapse than BP, but with much lower accuracy and no error bars.
Natural patterns of activity and long-term synaptic plasticity,
1 Pith paper cite this work, alongside 348 external citations. Polarity classification is still indexing.
1
Pith paper citing it
348
external citations · OpenAlex
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Energy-Efficient Information Representation in MNIST Classification Using Biologically Inspired Learning
On a 3-class MNIST subset, a competitive-Hebbian + weight-perturbation rule achieves higher estimated mutual information per nonsilent synapse than BP, but with much lower accuracy and no error bars.