Activity-Based Plasticity in Spiking Neuron Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Artificial spiking neural networks face challenges in maintaining responsiveness to stimuli that appear at long intervals, leading to unstable input synaptic sets and receptive fields due to diminished responsiveness over time.
Innovation Solution
Implementing activity-based plasticity mechanisms in spiking neuron networks that adjust connection efficacies based on response rates and neuron activity, allowing for potentiation or depression of connections depending on the timing and frequency of input stimuli.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional spiking neural networks are used without activity-based plasticity, then the network structure remains static and simple, but the network loses responsiveness to stimuli appearing at long intervals and exhibits unstable synaptic sets
Solution Approach 1:
The patent implements dynamic adjustment of synaptic connection efficacies based on neuron response rates. Connections are continuously modified through potentiation or depression mechanisms that adapt to changing input patterns, allowing the network to maintain responsiveness to infrequent stimuli while preserving stability through rate-modulated plasticity rules
Solution Approach 2:
The patent employs feedback mechanisms where the response rate of post-synaptic neurons is monitored and used to modulate the plasticity of incoming connections. This feedback loop ensures that connections providing relevant inputs are strengthened while maintaining overall network stability through regulated adaptation
2Adaptability or versatility
If connection efficacies are continuously adjusted to maintain responsiveness, then the network can detect infrequent stimuli, but the complexity of managing plasticity rules increases
Solution Approach 1:
The patent changes the operational parameters of synaptic connections dynamically based on measured response rates. By modulating connection efficacies as adjustable parameters in response to observed neural activity patterns, the network achieves high adaptability to varying stimulus frequencies while managing complexity through parameter-based control rather than structural changes
3Measurement precision
If spike timing dependent plasticity is used, then learning precision is improved, but the network becomes sensitive to timing variations and less robust
Solution Approach 1:
The patent transitions from timing-dependent to rate-dependent plasticity, changing the controlling parameter from precise spike timing intervals to average response rates. This parameter change maintains learning precision through rate modulation while significantly improving robustness by eliminating sensitivity to exact timing variations in spike arrivals
Data Source
AI summary
Apparatus and methods for activity based plasticity in a spiking neuron network adapted to process sensory input. In one approach, the plasticity mechanism of a connection may comprise a causal potentiation portion and an anti-causal portion. The anti-causal portion, corresponding to the input into a neuron occurring after the neuron response, may be configured based on the prior activity of the neuron. When the neuron is in low activity state, the connection, when active, may be potentiated by a base amount. When the neuron activity increases due to another input, the efficacy of the connection, if active, may be reduced proportionally to the neuron activity. Such functionality may enable the network to maintain strong, albeit inactive, connections available for use for extended intervals.