Neural Core Interconnect for Synaptic Plasticity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current neuromorphic and synaptronic computation systems lack effective implementation of synaptic, dendritic, somatic, and axonal plasticity in neural networks, limiting their ability to mimic biological brain functionality and efficiency.
Innovation Solution
A neural network architecture with multiple functional neural core circuits and a dynamically reconfigurable switch interconnect, utilizing connectivity neural core circuits for bidirectional information flow and synaptic weight learning, enabling structural and functional plasticity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional digital models are used for neuromorphic computation, then computational functionality is achieved, but biological brain functionality and efficiency are not effectively mimicked
Solution Approach 1:
The system segments neural network functionality into modular neural core circuits, each containing neurons, axons, and synapses as separate functional units. This segmentation enables independent implementation of biological plasticity mechanisms in each module while maintaining overall system functionality.
Solution Approach 2:
The neural core circuits are designed as universal building blocks that can perform multiple functions: synaptic plasticity, dendritic plasticity, somatic plasticity, and axonal plasticity. The same hardware structure supports different types of biological learning mechanisms through configurable connectivity and plasticity rules.
2Adaptability or versatility
If synaptic, dendritic, somatic, and axonal plasticity are implemented, then learning capabilities are enhanced, but hardware complexity increases
Solution Approach 1:
Multiple plasticity mechanisms (synaptic, dendritic, somatic, axonal) are merged into a unified neural core circuit architecture. The hardware structure combines different plasticity functions in an integrated manner, allowing simultaneous operation of multiple learning mechanisms without requiring separate dedicated hardware for each type.
Solution Approach 2:
The system implements dynamic reconfigurability where connectivity patterns and plasticity rules can be modified during operation. The hardware supports time-varying connections and adaptive learning rules that change based on computational requirements, enabling flexible implementation of different plasticity mechanisms.
3Adaptability or versatility
If dynamically reconfigurable switch interconnect is used, then structural and functional plasticity are enabled, but system complexity increases
Solution Approach 1:
The system implements partial reconfigurability rather than full reconfigurability of the interconnect. Specific portions of the switch fabric are made dynamically reconfigurable to support the required plasticity functions, while other portions remain fixed to reduce overall system complexity.
Data Source
AI summary
Embodiments of the invention provide a neural network comprising multiple functional neural core circuits, and a dynamically reconfigurable switch interconnect between the functional neural core circuits. The interconnect comprises multiple connectivity neural core circuits. Each functional neural core circuit comprises a first and a second core module. Each core module comprises a plurality of electronic neurons, a plurality of incoming electronic axons, and multiple electronic synapses interconnecting the incoming axons to the neurons. Each neuron has a corresponding outgoing electronic axon. In one embodiment, zero or more sets of connectivity neural core circuits interconnect outgoing axons in a functional neural core circuit to incoming axons in the same functional neural core circuit. In another embodiment, zero or more sets of connectivity neural core circuits interconnect outgoing and incoming axons in a functional neural core circuit to incoming and outgoing axons in a different functional neural core circuit, respectively.