Neural Network Decision Support System for Preference Modeling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing decision support systems require extensive interaction between decision-makers and model designers to elicit preference information, which is often hindered by limited time and noisy data, making it challenging to develop accurate preference models.

Innovation Solution

A method for generating a multiple-criteria decision support system using a neural network architecture with specific constraints and training techniques, allowing for the creation of a decision support system with minimal intervention from decision-makers, utilizing training data to determine optimal model parameters.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If extensive interaction between decision-makers and model designers is used to elicit preference information, then the accuracy of preference modeling is improved, but the time required and human intervention needed increase

Engineering Contradiction:

Improveaccuracy of preference modelingVSAvoidtime required for model development

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system uses automated machine learning algorithms to independently analyze training data and determine preference model parameters without requiring extensive manual elicitation from decision-makers. The system serves itself by automatically extracting preferences from historical data, reducing the need for human intervention in the model development process.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical process of manual question-and-answer interaction between designers and decision-makers with automated computational algorithms. Machine learning models automatically process training data to infer preference parameters, substituting the manual elicitation mechanism with an automated digital system.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If manual elicitation of preference information is used, then the quality of preference data is improved, but the complexity of the development process increases

Engineering Contradiction:

Improvequality of preference dataVSAvoidcomplexity of development process

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces complex manual processes for data collection and preference extraction with automated machine learning algorithms. The system automatically processes raw data, identifies patterns, and extracts preference information without requiring complex manual procedures, thereby reducing development process complexity while maintaining data quality.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces automated algorithms as intermediaries between raw training data and preference model parameters. These algorithms act as mediators that automatically transform unstructured data into structured preference information, simplifying the overall development process while ensuring data quality through systematic processing.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Extent of automation

If training data is used to determine preference model parameters, then the need for human intervention is reduced, but the reliability of the data may be compromised by noise and errors

Engineering Contradiction:

Improveneed for human interventionVSAvoidreliability of training data

Core Design Contradiction:

Extent of automationVSReliability

Solution Approach 1:

The patent implements feedback mechanisms where the system continuously monitors and validates the quality of extracted preference parameters against the original training data. This feedback loop allows the system to detect and correct inconsistencies, ensuring that automated parameter determination maintains high reliability despite the presence of noisy or erroneous data.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent applies data cleaning and validation techniques beforehand to cushion against the effects of noise and errors in training data. By pre-processing the data to remove obvious errors and inconsistencies, the system protects the reliability of the final preference model parameters before the automated extraction process begins.

Inventive Principle:

Principle #11Beforehand cushioning (Prior cushioning)

Data Source

PatentUS20230206036A1Method for generating a decision support system and associated systems

Publication Date: 2023.06.29 THALES SA

AI summary

The present invention relates to a method for generating a multiple-criteria decision support system comprising:providing a problem and training data solving the problem for specific cases, the problem being a problem of evaluating the quality of a system chosen from:choosing the best alternative from among alternatives,distributing alternatives among classes,the storage of alternatives in order of preference, andproviding a score of an alternative,re-transcribing the problem according to a neural network and constraints to be observed,training the re-transcribed neural network using the training data,the determination of the function performed by the trained neural network, andphysically implementing the determined function in order to obtain the support system.