Counterfactual Explanation Clustering for AI Feedback Efficiency

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Automated analysis systems often provide infeasible counterfactual explanations, wasting computing and communication resources by processing and delivering irrelevant scenarios, which degrades user experience and inefficiently uses feedback for system updates.

Innovation Solution

The system receives user information and prediction outputs, generates and clusters counterfactual explanations, selects relevant ones based on relevance scores, requests feedback, updates models, and retrains using labeled explanations to determine optimal counterfactuals, conserving resources and improving explanation relevance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If the system provides multiple counterfactual explanations to users, then the completeness of information is improved, but the computing and communication resources are wasted due to processing and delivering irrelevant scenarios

Engineering Contradiction:

Improvecompleteness of counterfactual explanationsVSAvoidcomputing and communication resources

Core Design Contradiction:

Loss of informationVSLoss of energy

Solution Approach 1:

The patent extracts and removes irrelevant counterfactual explanations from the set of generated explanations before presenting them to users. The system identifies and discards explanations that do not meet relevance criteria, thereby reducing resource consumption while maintaining the quality and usefulness of the remaining explanations.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system generates an excessive number of counterfactual explanations initially (to ensure completeness) but then applies filtering mechanisms to retain only the necessary subset. This partial action approach allows the system to explore multiple possibilities while ultimately delivering only the most relevant ones to users.

Inventive Principle:

Principle #16Partial or excessive action

2Quantity of substance

If the system provides infeasible counterfactual explanations, then the quantity of explanations is increased, but the user experience is degraded due to irrelevant scenarios

Engineering Contradiction:

Improvequantity of counterfactual explanationsVSAvoiduser experience

Core Design Contradiction:

Quantity of substanceVSEase of operation

Solution Approach 1:

The patent converts the potential harm of providing infeasible explanations into a benefit by using the filtering process to identify and eliminate irrelevant scenarios. The system leverages the initial generation of multiple explanations (including infeasible ones) as a means to train and improve its relevance assessment models, ultimately delivering higher quality explanations.

Inventive Principle:

Principle #22Blessing in disguise (Convert harm into benefit)

3Loss of information

If the system processes and delivers all generated counterfactual explanations, then the comprehensiveness is improved, but the feedback efficiency is reduced due to processing irrelevant explanations

Engineering Contradiction:

Improvecomprehensiveness of explanationsVSAvoidfeedback processing efficiency

Core Design Contradiction:

Loss of informationVSProductivity

Solution Approach 1:

The system performs preliminary filtering of counterfactual explanations before they are delivered to users. By pre-assessing relevance and eliminating irrelevant explanations in advance, the system reduces the burden of processing feedback on infeasible scenarios, thereby improving overall feedback efficiency while maintaining comprehensiveness of the delivered explanations.

Inventive Principle:

Principle #10Preliminary action

4Loss of information

If the system provides detailed counterfactual explanations, then the information quality is improved, but the communication resources are wasted due to delivering irrelevant scenarios

Engineering Contradiction:

Improvequality of explanationsVSAvoidcommunication resources

Core Design Contradiction:

Loss of informationVSLoss of energy

Solution Approach 1:

The patent applies local quality by providing detailed and comprehensive explanations only for the most relevant counterfactual scenarios, while summarizing or omitting details for less relevant ones. This differentiated approach ensures high information quality where it matters most while reducing communication resource consumption for peripheral explanations.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12254388B2Generation of counterfactual explanations using artificial intelligence and machine learning techniques

Publication Date: 2025.03.18 ACCENTURE GLOBAL SOLUTIONS LTD

AI summary

In some implementations, a system may determine, based on a qualification model, a prediction output of an analysis of user information. The system may determine, based on a generator model, a plurality of counterfactual explanations associated with the prediction output and the user information. The system may cluster, according to a clustering model, the plurality of counterfactual explanations into clusters of counterfactual explanations. The system may select, based on a classification model, a counterfactual explanation from a cluster of the clusters of counterfactual explanations. The system may provide a request for feedback associated with the counterfactual explanation. The system may receive feedback data associated with the request for feedback. The system may update a data structure associated with the clustering model based on the feedback data and the counterfactual explanation to form an updated data structure. The system may perform an action associated with the updated data structure.