Adaptive Plasticity in Spiking Neuron Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing artificial spiking neural networks face challenges in adapting to varying input characteristics and dynamic regimes, limiting their ability to effectively learn and recognize features in a wide range of conditions.
Innovation Solution
The implementation of adaptive plasticity mechanisms in spiking neuron networks, which adjust connection efficacies based on similarity measures between neuron outputs and inputs, incorporating inhibitory connections to delay or prevent post-synaptic responses, thereby enhancing learning and feature recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing plasticity implementations are used, then network learning can be achieved, but the network cannot adapt to varying input characteristics and dynamic regimes
Solution Approach 1:
The patent implements dynamic plasticity mechanisms where connection efficacies are continuously adjusted based on real-time similarity measures between neuron outputs and inputs. The system transitions from static weight adjustments to dynamic, condition-dependent plasticity rules that adapt to varying input characteristics and network states, enabling the network to effectively learn across diverse dynamic regimes
Solution Approach 2:
The patent changes the parameters of plasticity rules based on similarity measures computed from neuron activity patterns. By modulating plasticity strength and direction according to these similarity parameters, the system enables adaptive learning that responds to varying input characteristics while maintaining reliable feature recognition across different conditions
2Adaptability or versatility
If more neurons are used to recognize richer feature sets, then feature recognition capability improves, but network complexity and resource requirements increase
Solution Approach 1:
The patent modifies connection efficacies dynamically based on similarity measures, allowing existing neurons to develop diverse effective receptive fields through parameter adaptation rather than requiring additional neurons. This enables richer feature recognition capability while maintaining or reducing the physical neuron count through intelligent parameter modulation
Data Source
AI summary
Apparatus and methods for modulated plasticity in a spiking neuron network. A plasticity mechanism may be configured for example based on a similarity measure between post-synaptic activities of two or more neurons that may be receiving the same feed-forward input. The similarity measure may comprise a dynamically determined cross-correlogram between the output spike trains of two neurons. An a priori configured similarity measure may be used during network operation in order to update efficacy of inhibitory connections between neighboring neurons. Correlated output activity may cause one neuron to inhibit output generation by another neuron thereby hindering responses by multiple neurons to the same input stimuli. The inhibition may be based on an increased efficacy of inhibitory lateral connection. The inhibition may comprise modulation of the pre synaptic portion the plasticity rule based on efficacies of feed-forward connection and inhibitory connections and a statistical parameter associated with the post-synaptic rule.