Defective-Input Prediction Evaluation Using Complementary Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to evaluate and suggest effective changes to attribute values when input data has defects, making it difficult to achieve desired prediction results.
Innovation Solution
The system generates complementary data patterns to address defects, determines perturbation vectors for attribute changes, and evaluates these vectors based on determination results and cost values to suggest actionable changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complementary data is generated by complementing defects in multiple ways, then the ability to evaluate perturbation information improves, but the computational complexity and processing time increase
Solution Approach 1:
The system pre-generates multiple complementary data patterns by complementing defective input data in various ways before perturbation evaluation. This preliminary generation of diverse data patterns enables more accurate evaluation of whether perturbation information can change labels across different patterns, resolving the contradiction by preparing evaluation candidates in advance rather than computing them on-demand during evaluation
Solution Approach 2:
The evaluation process is segmented into independent assessments for each complementary data pattern. The system evaluates whether perturbation information changes labels in each pattern separately, then aggregates these determination results. This segmentation allows parallel processing of different patterns, improving evaluation accuracy while managing computational complexity through modular assessment
2Reliability
If multiple complementary data patterns are generated to evaluate perturbation information, then the reliability of prediction result changes improves, but the device complexity increases
Solution Approach 1:
The complementary data generation mechanism serves multiple functions: it handles various types of data defects, generates diverse evaluation patterns, and enables robust perturbation assessment. By making the data complementation system universal and multi-functional, the patent improves prediction reliability across different defect scenarios without proportionally increasing system complexity
Solution Approach 2:
Complementary data patterns act as intermediaries between defective input data and perturbation evaluation. Instead of directly evaluating perturbations on incomplete data, the system introduces multiple complemented data patterns as intermediate representations. This intermediary approach enhances reliability by providing multiple evaluation contexts while keeping the core evaluation logic relatively simple
Data Source
AI summary
A computer-readable recording medium has stored therein an evaluation program causing a computer to execute a process. The process includes: generating, when a portion of values of a plurality of attributes included in input data has a defect, complementary data of a plurality of patterns obtained by complementing the defect in a plurality of ways; determining perturbation information including an attribute to be changed and a change amount from among the plurality of attributes of complementary data in order to change a label predicted by the complementary data of the plurality of patterns; and evaluating the perturbation information based on a determination result as to whether the perturbation information determined in the complementary data of one pattern among the plurality of patterns can change the label also with respect to the complementary data of another pattern among the plurality of patterns.