Data Exchange Prediction Model for Reliable Entity Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for data exchange between entities in the naval, maintenance, and medical fields are inefficient and lack reliability, particularly in identifying inconsistencies or anomalies, which can lead to time-consuming monitoring and uncertain results.
Innovation Solution
A data exchange method utilizing a prediction model to classify elements based on descriptor vectors, with the ability to generate counterfactual examples to better understand prediction outputs, improving the reliability and interpretability of data analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual monitoring is implemented to detect inconsistent routes, then detection capability is improved, but time consumption increases
Solution Approach 1:
The patent replaces manual monitoring operations with an automated prediction model that processes descriptor vectors and generates predictions automatically. The system substitutes human operators with computational algorithms that can analyze data without time constraints, eliminating the trade-off between detection capability and time consumption.
Solution Approach 2:
The system enables self-service through automated prediction models that independently analyze data and generate results without requiring manual intervention. The prediction model processes incoming descriptor vectors, generates predictions, and provides explanations automatically, allowing the system to serve itself without external human resources.
2Loss of information
If in-depth monitoring is implemented to understand unusual routes, then understanding depth is improved, but result reliability deteriorates
Solution Approach 1:
The patent implements feedback through explanation generation that provides insights into prediction reasoning. The system analyzes the prediction process and generates explanations that feed back to operators, improving understanding while maintaining reliability through transparent, traceable prediction logic rather than opaque deep monitoring.
Solution Approach 2:
The prediction model acts as an intermediary between raw data and human interpretation. It processes descriptor vectors through a structured prediction framework and provides both predictions and explanations, serving as a reliable mediator that maintains result reliability while enhancing understanding depth.
3Productivity
If prediction models are used to classify elements, then analysis speed is improved, but interpretability deteriorates
Solution Approach 1:
The patent applies preliminary action by generating explanations alongside predictions in the same processing step. Rather than adding interpretation later, the system prepares both the prediction result and its explanation simultaneously, maintaining analysis speed while ensuring interpretability is preserved from the outset.
Solution Approach 2:
The system adds another dimension to the output by providing explanations that complement predictions. This transforms the output from a single-dimensional prediction to a two-dimensional result including both prediction and explanation, enhancing interpretability without sacrificing analysis speed.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to a method for exchanging data between entities (6, 7, 8), the method comprising the steps of: - receiving a descriptor vector of a first element, - predicting the class of the first element using a prediction model on the basis of the descriptor vector of the first element, - determining the contribution of each variable describing the first element in order to ascertain the class of the first element, - selecting a subset of variables describing the first element, - determining, from at least a second element on the basis of the descriptor vector of the first element, the class predicted for the first element and the selected subset of variables describing the first element, and - sending the descriptor vector of the first element, the class of the first element, the descriptor vector of the second element and the class of the second element.