Dual-Level Input Processing for Adaptive AI Classification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing artificial intelligence systems, particularly neural networks, are typically designed for specialized tasks and struggle to adapt to different domains without extensive retraining, leading to inefficiencies and the inability to provide real-time responses in high-dimensional spaces.

Innovation Solution

A method involving a working level and an evaluation level of artificially learning systems, where input values are processed through a first and second classification, with the evaluation level influencing the working level to meet predetermined conditions, incorporating non-technical aspects like moral and ethical considerations, and using modulation functions to adjust neural network parameters.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If neural networks are designed for specialized tasks with fixed architectures, then they achieve high accuracy in specific domains, but they cannot adapt to different domains without extensive retraining

Engineering Contradiction:
Improveclassification accuracyVSAvoiddomain adaptability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent implements dynamic neural networks where the architecture can be modified at runtime through the addition, removal, or modification of nodes and connections. This allows the network to adapt its structure to different classification tasks and domains without complete retraining, resolving the contradiction between specialized accuracy and general adaptability

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The neural network is divided into independent, modifiable components (nodes and connections) that can be individually added, removed, or adjusted. This segmentation enables flexible reconfiguration of the network architecture for different domains while preserving useful existing structures, balancing specialization and adaptability

Inventive Principle:
Principle #1Segmentation

2Adaptability or versatility

If neural networks process high-dimensional input spaces, then they achieve comprehensive classification coverage, but real-time response becomes impossible due to computational complexity

Engineering Contradiction:
Improveclassification coverageVSAvoidreal-time response speed
Core Design Contradiction:
Adaptability or versatilityVSSpeed

Solution Approach 1:

The patent pre-calculates and stores distance metrics and similarity measures between data points during an offline phase. This preliminary computation allows the online classification phase to operate rapidly by simply comparing pre-computed values rather than performing complex high-dimensional calculations in real-time

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system computes classification results for a subset of relevant neighbors or data points rather than processing the entire high-dimensional space. By focusing computational effort on the most relevant portions of the data space, the system achieves real-time response while maintaining comprehensive classification coverage

Inventive Principle:
Principle #16Partial or excessive action

3Reliability

If AI systems are trained extensively for specialized tasks, then they achieve high performance, but they become useless in other domains and require complete retraining

Engineering Contradiction:
Improvetask performanceVSAvoidcross-domain usability
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system incorporates feedback mechanisms that allow it to learn from classification results and adjust its architecture accordingly. Performance feedback triggers automatic modifications to the neural network structure, enabling the system to maintain high reliability across different domains by continuously adapting to new tasks while preserving useful learned patterns

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4248284B1Method and system for processing input values
Publication Date: 2025.10.01 FRAUNHOFER GESELLSCHAFT ZUR FORDERUNG DER ANGEWANDTEN FORSCHUNG EV
  • EP4248284B1 patent drawingFigure 1
  • EP4248284B1 patent drawingFigure 2a~2h
  • EP4248284B1 patent drawingFigure 3

AI summary

The invention relates to a method implemented in a controller of a machine for processing input values in an overall system having a working level and a rating level, comprising inputting first input values to the working level and determining first output values; forming first situation data; inputting the first situation data to the rating level and determining first ratings that indicate whether the first situation data meet predefined first conditions; influencing the determination of the first output values based on the first ratings; inputting second input values to the working level and determining second output values, wherein the determination of the second output values is influenced by the first output values; forming second situation data; inputting the second situation data to the rating level and determining second ratings that indicate whether the second situation data meet predefined second conditions, wherein the determination of the second ratings is influenced by the first ratings; influencing the determination of the second output values based on the second ratings; wherein the first and/or the second output values are used as overall output values of the overall system.