Neural Pattern Replay via Referencing Neurons
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for neural system engineering fail to truly replay neural firing patterns learned by neurons in the absence of the original stimulus, and lack effective solutions for fast learning, learning refinement, association, and memory transfer after the stimulus is no longer present.
Innovation Solution
A method and apparatus for neural component replay that reference patterns in afferent neuron outputs, match relational aspects between these outputs and referencing neurons, and induce afferent neurons to output a substantially similar pattern, enabling true replay, fast learning refinement, and memory transfer without the original stimulus.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If current methods are used to determine what pattern a neuron matches, then the method can identify patterns, but it requires trying different patterns until the matching one is found, which is inefficient
Solution Approach 1:
The patent creates a copy of the neural pattern through the referencing neuron system. When a neuron fires a pattern, referencing neurons capture and store this pattern, allowing the system to replay the exact same pattern without needing to re-stimulate the original neuron or try different patterns. This copying mechanism enables direct pattern retrieval and comparison, dramatically improving identification efficiency.
Solution Approach 2:
The patent performs preliminary action by having referencing neurons capture and store neural patterns during the initial firing event. Instead of waiting to determine what pattern was fired, the system proactively records the pattern in the referencing neurons while it is being produced, making it immediately available for future identification and comparison without requiring additional time-consuming operations.
2Adaptability or versatility
If the original stimulus is removed after neural pattern learning, then the system can operate independently, but it cannot truly replay the learned neural firing pattern
Solution Approach 1:
The patent introduces referencing neurons as intermediaries between the original stimulus-responsive neurons and the pattern replay function. These referencing neurons capture the neural pattern during stimulation and maintain it as a stored representation. When the original stimulus is removed, the referencing neurons serve as the mediator that enables faithful pattern replay by reactivating the same neural sequence without requiring the original stimulus or neuron to be present.
3Productivity
If functional one-way learning methods are used, then the system can learn patterns, but it cannot efficiently perform learning refinement, association, and memory transfer after the original stimulus is gone
Solution Approach 1:
The patent implements feedback by having the referencing neurons send signals back to the original neurons whose patterns they referenced. This feedback loop enables learning refinement because the system can replay stored patterns and use them to strengthen or modify synaptic connections. The referencing neurons provide a feedback mechanism that allows the system to review and refine learned patterns without requiring the original stimulus, enabling continuous learning improvement.
Solution Approach 2:
The patent makes the referencing neuron system universal by enabling it to perform multiple functions: capturing patterns during learning, storing patterns for later retrieval, replaying patterns independently, refining learning through feedback, associating patterns through hierarchical relationships, and transferring memory between neural populations. This multi-functional system replaces the need for separate mechanisms for each learning operation.
Data Source
AI summary
Aspects of the present disclosure support techniques for neural pattern sequence completion and neural pattern hierarchical replay. At least a portion of a pattern can be invoked for replay upon referencing the pattern and learning relational aspects between elements of the pattern and the referencing of the pattern using hierarchical levels of neurons.