Inverted STDP Feedback Stabilizes Spiking Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing spiking neural networks face instability due to runaway positive feedback loops, leading to uncontrolled spiking and hallucinatory responses, as they lack effective mechanisms for stable feedback implementation.
Innovation Solution
The implementation of inverted spike-timing dependent plasticity (STDP) for feedback connections, where weights associated with context connections are depressed when the context signal precedes the spike and potentiated when the spike precedes the context signal, to control and stabilize the network operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feedback connections are implemented in spiking neural networks, then detection capabilities and sensory input representation are enhanced, but network stability deteriorates due to runaway positive feedback loops
Solution Approach 1:
The patent inverts the conventional STDP rule for feedback connections. Instead of potentiating synapses when pre-synaptic spikes precede post-synaptic spikes (standard STDP), the patent applies depression when context signals precede spikes and potentiation when spikes precede context signals. This inverted timing rule fundamentally reverses the feedback mechanism to prevent runaway excitation while maintaining detection capabilities.
Solution Approach 2:
The patent implements a feedback mechanism where the timing relationship between context signals and spiking activity dynamically adjusts synaptic weights. This feedback loop continuously monitors the temporal correlation between contextual information and neuronal responses, automatically strengthening or weakening connections based on whether context precedes or follows the spike, thereby stabilizing network operation.
2Manufacturing precision
If context connections are strengthened to improve sensory representation, then accurate representation of sensory inputs is enhanced, but uncontrolled spiking occurs due to positive feedback loops
Solution Approach 1:
The patent applies inverted STDP specifically to context connections, reversing the conventional potentiation-depression pattern. When context signals precede spikes (which would normally strengthen the connection in standard STDP), the patent instead depresses the connection. This prevents context-driven runaway spiking while preserving the ability to accurately represent sensory inputs through selective potentiation when spikes precede context signals.
Solution Approach 2:
The patent implements preliminary anti-action by preemptively depressing context connections when context signals are detected to precede spiking activity. This counter-measure is applied before runaway positive feedback can develop, preventing the formation of unstable feedback loops that would lead to uncontrolled spiking and hallucinatory responses.
Data Source
AI summary
Apparatus and methods for feedback in a spiking neural network. In one approach, spiking neurons receive sensory stimulus and context signal that correspond to the same context. When the stimulus provides sufficient excitation, neurons generate response. Context connections are adjusted according to inverse spike-timing dependent plasticity. When the context signal precedes the post synaptic spike, context synaptic connections are depressed. Conversely, whenever the context signal follows the post synaptic spike, the connections are potentiated. The inverse STDP connection adjustment ensures precise control of feedback-induced firing, eliminates runaway positive feedback loops, enables self-stabilizing network operation. In another aspect of the invention, the connection adjustment methodology facilitates robust context switching when processing visual information. When a context (such an object) becomes intermittently absent, prior context connection potentiation enables firing for a period of time. If the object remains absent, the connection becomes depressed thereby preventing further firing.