Recurrent Neural Network Clique Patterns for Decision Moment Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing data processing systems, including traditional computers, rely on predefined logic sequences to determine decision moments, making it difficult to identify when information processing is complete, whereas artificial recurrent neural networks lack a clear identifiable decision point due to their dynamic and cyclical nature.

Innovation Solution

Characterize the dynamic properties of recurrent artificial neural networks by identifying clique patterns and directed clique patterns in the network activity, using methods to determine decision moments based on the timing and complexity of these patterns, which can be represented as binary sequences for digital processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional computers use predefined logic sequences to process information, then decision moments are easy to identify, but the system lacks adaptability to dynamic patterns

Engineering Contradiction:

Improvedecision moment identificationVSAvoidadaptability to dynamic patterns

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent applies dynamics by transitioning from static predefined logic sequences to dynamic pattern recognition. The system continuously monitors network activity and identifies decision moments based on real-time complexity measurements of activity patterns, allowing the system to adapt to changing dynamics in recurrent neural networks while maintaining precise decision moment identification

Inventive Principle:

Principle #15Dynamics

2Adaptability or versatility

If recurrent neural networks operate with dynamic cyclical nature, then adaptability is improved, but decision moments become difficult to identify

Engineering Contradiction:

Improvedynamic processing capabilityVSAvoiddecision moment identification

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent implements feedback by measuring the complexity of network activity patterns and using this information to identify decision moments. The system continuously monitors activity complexity, compares it against thresholds or historical patterns, and uses this feedback to pinpoint decision moments within the dynamic cyclical operation of recurrent neural networks, resolving the identification difficulty while preserving adaptability

Inventive Principle:

Principle #23Feedback

3Productivity

If activity patterns are characterized by high complexity, then information processing capability is improved, but the difficulty of detecting and measuring increases

Engineering Contradiction:

Improveinformation processing capabilityVSAvoidactivity pattern analysis

Core Design Contradiction:

ProductivityVSDifficulty of detecting and measuring

Solution Approach 1:

The patent introduces complexity metrics as an intermediary that bridges high-complexity activity patterns and measurable decision moment identification. Instead of directly analyzing complex activity patterns, the system uses complexity measurements as a mediator to translate intricate network dynamics into quantifiable data that can be used to identify decision moments, making high-capacity processing measurable and manageable

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS12412072B2Characterizing activity in a recurrent artificial neural network

Publication Date: 2025.09.09 INAIT SA

AI summary

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for characterizing activity in a recurrent artificial neural network. In one aspect, a method includes outputting digits from a recurrent artificial neural network, wherein each digit represents whether or not activity within a particular group of nodes in the recurrent artificial neural network comports with a respective pattern of activity.