Genetic Programming for Image Classification Program Generation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image classification techniques require advanced knowledge and significant development time to optimize the combination of feature values and classifiers, and existing methods using evolutionary computation do not include the generation of classifiers as a target for optimization.

Innovation Solution

A program generating apparatus that uses genetic programming to optimize the combination of image filters and classifier generating algorithms by generating tree structures with a fitness score exceeding a predetermined threshold, thereby producing highly accurate image classification programs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If evolutionary computation is used to optimize the filter bank, then the automation of image processing program generation is improved, but the classification accuracy is insufficient because classifier generation is not included in the optimization target

Engineering Contradiction:

Improveautomation of image processing program generationVSAvoidclassification accuracy

Core Design Contradiction:

Extent of automationVSMeasurement precision

Solution Approach 1:

The patent combines the optimization of the filter bank and the generation of the classifier into a single integrated process. The program generation apparatus simultaneously optimizes both the preprocessing filter bank and the classifier generation algorithm, allowing them to work together as a unified system rather than separate components.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The program generation apparatus performs multiple functions: it generates the filter bank, generates the classifier, and optimizes both simultaneously using genetic programming. This multi-functional approach allows a single system to handle both preprocessing and classification generation, improving overall automation while maintaining high accuracy.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Measurement precision

If experts manually optimize the combination of feature values and classifiers, then the classification accuracy is improved, but the development time and complexity increase significantly

Engineering Contradiction:

Improveclassification accuracyVSAvoiddevelopment time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs self-optimization through genetic programming, automatically generating and optimizing both the filter bank and classifier without requiring expert manual intervention. The genetic programming algorithm evolves the program structure autonomously, allowing the system to improve its own classification capability while reducing dependency on expert knowledge.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent uses genetic programming to automatically adjust and optimize program parameters and structure. The system changes parameters such as filter types, feature extraction methods, and classifier configurations through evolutionary computation, finding optimal combinations without manual expert tuning.

Inventive Principle:

Principle #35Parameter changes

3Manufacturing precision

If genetic programming optimizes only the filter bank, then the preprocessing is improved, but the overall classification performance is limited due to lack of classifier optimization

Engineering Contradiction:

Improvepreprocessing qualityVSAvoidoverall classification performance

Core Design Contradiction:

Manufacturing precisionVSMeasurement precision

Solution Approach 1:

The patent merges the optimization of the filter bank with the generation and optimization of the classifier into a single genetic programming process. Both components are optimized simultaneously, ensuring that the preprocessing and classification work together effectively rather than in isolation.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The system performs preliminary generation of both the filter bank and classifier through genetic programming before actual classification. The genetic programming process pre-optimizes the entire pipeline, including both preprocessing and classification components, preparing them for high-performance classification tasks.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11061687B2Apparatus and method for program generation

Publication Date: 2021.07.13 FUJITSU LTD

AI summary

There is provided a program generating apparatus including a generating unit and a genetic processing unit. The generating unit is configured to generate tree structures each representing an image classification program. Each of the tree structures has a first level group and a second level group. Elements of nodes in the first level group are selected from amongst image filters each used to apply preprocessing to an input image. An element of a node in the second level group is selected from amongst setting programs each used to set a different value as a control parameter for generating a classifier based on information obtained by execution of the elements selected for the nodes in the first level group. The genetic processing unit is configured to output, using genetic programming, a tree structure with a fitness score exceeding a predetermined threshold based on the tree structures.