Neural Network Node Assemblies for Information Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neural network technologies face challenges in organizing nodes effectively to achieve improved information processing and storage, as they often lack the structural organization and connectivity patterns found in biological neurons, leading to suboptimal performance in non-linear data processing tasks.
Innovation Solution
A neural network device is implemented with node assemblies interconnected by links, where the strength of connections is determined by the number of common neighbors, mimicking the organization of biological neurons in the rat neocortex, allowing for improved training and information processing capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If nodes are organized with higher within-assembly connectivity than between-assembly connectivity, then information processing and storage performance is improved, but device complexity increases due to the structured organization requirements
Solution Approach 1:
The neural network is divided into multiple node assemblies, where each assembly contains a subset of nodes that are more densely interconnected within the assembly than with nodes in other assemblies. This segmentation creates a modular structure that improves information processing while managing complexity through organized subunits.
Solution Approach 2:
Different regions of the network (node assemblies) are given different connectivity characteristics. Within each assembly, nodes have high connectivity to each other, while between assemblies, connectivity is lower. This local quality differentiation optimizes processing within regions while maintaining overall network functionality.
2Adaptability or versatility
If connection weights are adjusted to reflect biological neuron organization patterns, then training effectiveness is improved, but manufacturing precision requirements increase
Solution Approach 1:
The connection weights between nodes are adjusted based on the number of common neighbors shared by connected nodes. Specifically, weights are set proportional to the number of common neighbors, creating a parameter-based organization that mimics biological neural connectivity patterns and improves training effectiveness.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for organizing trained and untrained neural networks. In one aspect, a neural network device includes a collection of node assemblies interconnected by between-assembly links, each node assembly itself comprising a network of nodes interconnected by a plurality of within-assembly links, wherein each of the between-assembly links and the within-assembly links have an associated weight, each weight embodying a strength of connection between the nodes joined by the associated link, the nodes within each assembly being more likely to be connected to other nodes within that assembly than to be connected to nodes within others of the node assemblies.