Non-intersecting Separation Surfaces for Reliable Data Classification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional data classification methods using separation surfaces struggle to simultaneously perform identification and outlying value classification with high reliability, as they often rely on single separation surfaces, neglecting boundaries on the opposite side, and fail to consider relationships between classes, leading to optimistic classification and reduced accuracy.

Innovation Solution

A data classification method and device that utilize a set of non-intersecting separation surfaces to define boundaries for both known and unknown classes, allowing for simultaneous identification and outlying value classification by calculating the classification target data's region in a feature space separated into known and unknown class regions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a single separation surface is used for data classification, then the classification process is simple and fast, but the reliability and accuracy of classification deteriorates due to neglecting boundaries on the opposite side and optimistic classification

Engineering Contradiction:
Improveclassification reliabilityVSAvoidclassification structure complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies segmentation by dividing the feature space into multiple non-overlapping regions using multiple separation surfaces. Each separation surface defines a boundary between different class regions, allowing the system to capture complex decision boundaries while maintaining computational efficiency through the non-intersecting property of the surfaces.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If multiple separation surfaces are used to define class boundaries, then classification accuracy and reliability improve, but the complexity of the classification system increases

Engineering Contradiction:
Improveclassification accuracyVSAvoidseparation surface set complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the feature space into distinct non-overlapping regions using multiple separation surfaces, where each surface contributes to defining class boundaries. This segmentation approach enables accurate classification by capturing complex boundaries while the non-intersecting constraint keeps the system manageable.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Each separation surface is optimized locally to define specific boundaries between classes, allowing the system to achieve high classification accuracy by focusing computational resources on local boundary definitions rather than attempting to optimize a single global surface.

Inventive Principle:
Principle #3Local quality

3Productivity

If conventional separation surface methods are used, then the classification process is computationally efficient, but the ability to simultaneously perform identification and outlying value classification deteriorates

Engineering Contradiction:
Improveclassification processing efficiencyVSAvoidmulti-function classification capability
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent implements multi-functionality by designing a classification system that can simultaneously perform both identification (classifying known classes) and outlying value detection (identifying anomalies) using the same set of multiple separation surfaces. The non-overlapping region structure naturally supports both functions without requiring separate processing pipelines.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS8589397B2Data classification method and data classification device
Publication Date: 2013.11.19 NEC CORP
  • US8589397B2 patent drawing
  • US8589397B2 patent drawing
  • US8589397B2 patent drawing

AI summary

A separation surface set storage part stores information defining a plurality of separation surfaces which separate a feature space into at least one known class region respectively corresponding to at least one known class and an unknown class region. Each of the at least one known class region is separated from outside region by more than one of the plurality of separation surfaces which do not intersect to each other. A data classification apparatus determine a classification of a classification target data whose inner product in the feature space is calculable by calculating to which region of the at least one known class region and the unknown class region determined by the information stored in the separation surface set storage part the classification target data belongs. A method and apparatus for data classification which can simultaneously perform identification and outlying value classification with high reliability in a same procedure are provided.