Spiking Neuron Network Saliency Detection via Adaptive Plasticity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing artificial spiking neural networks face challenges in efficiently identifying salient features within a wide variety of inputs due to computationally intensive processes, particularly in real-time visual processing, especially in portable devices.
Innovation Solution
Implementing adaptive plasticity mechanisms in spiking neuron networks that dynamically adjust connection activity, using inhibitory connections to enhance feature identification by differentiating and suppressing responses based on specific rules and efficacies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional spiking neural networks are used for saliency detection, then feature identification capability is maintained, but computational power requirements increase and processing speed decreases
Solution Approach 1:
The network is segmented into excitatory and inhibitory neuron populations with distinct functional roles. Excitatory neurons process feature detection while inhibitory neurons perform suppression, dividing the computational workload and enabling parallel processing that reduces overall computational requirements and increases processing speed.
Solution Approach 2:
The network implements dynamic plasticity mechanisms where connection weights adapt based on spike timing and activity patterns. This dynamic adjustment allows the network to optimize its computational resources in real-time, maintaining high processing speed while reducing energy consumption by only activating necessary computational pathways.
2Adaptability or versatility
If adaptive plasticity mechanisms are implemented, then connection activity adjustment capability is improved, but network complexity increases
Solution Approach 1:
Plasticity mechanisms are applied locally to specific connection types rather than uniformly across the entire network. Different plasticity rules are applied to excitatory-excitatory, excitatory-inhibitory, and inhibitory-excitatory connections, allowing adaptability where needed while maintaining simplicity in other portions of the network.
Solution Approach 2:
The network changes its operational parameters dynamically based on input characteristics. Plasticity modifiers adjust connection strengths based on spike timing differences and activity patterns, enabling the network to adapt its complexity level to match the demands of the specific processing task.
3Measurement precision
If inhibitory connections are used for feature differentiation, then saliency detection accuracy is improved, but computational overhead increases
Solution Approach 1:
The network merges feature detection and suppression functions into a single integrated processing stage. Excitatory neurons detect features while simultaneously receiving inhibitory input that suppresses non-salient features, combining what would otherwise be separate processing steps into one efficient operation that reduces computational overhead.
Solution Approach 2:
The inhibitory connections perform self-regulation through activity-dependent plasticity, automatically adjusting their suppression strength based on network activity patterns. This self-service mechanism eliminates the need for external control mechanisms, reducing computational overhead while maintaining high detection accuracy.
Data Source
AI summary
Apparatus and methods for salient feature detection by a spiking neuron network. The network may comprise feature-specific units capable of responding to different objects (red and green color). The plasticity mechanism of the network may be configured based on difference between two similarity measures related to activity of different unit types obtained during network training. One similarity measure may be based on activity of units of the same type (red). Another similarity measure may be based on activity of units of one type (red) and another type (green). Similarity measures may comprise a cross-correlogram and/or mutual information determined over an activity window. During network operation, the activity based plasticity mechanism may be used to potentiate connections between units of the same type (red-red). The plasticity mechanism may be used to depress connections between units of different types (red-green). The plasticity mechanism may effectuate detection of salient features in the input.