Genetic Programming Flow Generation Using Metadata Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining optimal processing flows, such as genetic programming, are inefficient due to the need for extensive crossover and mutation processes, especially when dealing with a large number of initial combination candidates, leading to prolonged search times.
Innovation Solution
A flow generation program and method that utilizes a non-transitory computer-readable storage medium to select processing flows with similar metadata, optimizing tree structures using genetic programming with a reduced number of initial individuals, thereby accelerating the search for optimal processing flows by referencing a database of pre-accumulated metadata and processing flows.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If genetic programming is used to search for optimal processing procedures, then the ability to find optimal solutions is improved, but the search time is excessively prolonged
Solution Approach 1:
The patent applies preliminary action by pre-generating a comprehensive library of candidate processing procedures (individuals) with various tree structures before the actual optimization search. This library is created in advance and stored for reuse, eliminating the need to generate individuals during the genetic programming execution, thus significantly reducing search time while maintaining the ability to find optimal solutions.
Solution Approach 2:
The patent uses copying by creating multiple copies of pre-generated processing procedure individuals and storing them in a library. These copied individuals can be repeatedly used across different genetic programming runs and problems, avoiding redundant generation work and accelerating the optimization process while preserving solution quality.
2Adaptability or versatility
If the number of initial combination candidates is increased, then the comprehensiveness of search is improved, but the processing time is excessively prolonged
Solution Approach 1:
The patent performs the computationally intensive generation of diverse candidate individuals in advance, creating a comprehensive library that covers a wide range of processing procedures. This preliminary generation allows the system to maintain high comprehensiveness while achieving fast processing speeds during actual use, as the library can be quickly queried and utilized without regenerating individuals.
Solution Approach 2:
By copying and storing numerous pre-generated individuals in a library, the system maintains a large and diverse set of candidates available for search without incurring the time cost of generating them during execution. This copying approach enables comprehensive coverage of the solution space while preserving high processing productivity.
3Measurement precision
If expert knowledge is required for processing flow determination, then the accuracy of flow selection is improved, but the difficulty of operation is excessively increased
Solution Approach 1:
The patent implements self-service by enabling the system to automatically generate, evaluate, and select optimal processing procedures without requiring expert human intervention. The genetic programming algorithm autonomously searches the pre-generated library, automatically determines the optimal flow, and produces results that would otherwise require expert knowledge, thereby making the system easy to operate while maintaining high accuracy.
Solution Approach 2:
The patent replaces the mechanical system of expert human judgment with an automated computational system. Instead of relying on experts to manually determine optimal processing flows, the system uses genetic programming algorithms to automatically search and select the best procedures from the pre-generated library, substituting human expertise with machine intelligence to achieve both ease of operation and accurate results.
Data Source
AI summary
A non-transitory computer-readable storage medium storing a flow generation program that causes at least one computer to execute a process, the process includes based on first metadata that corresponds to a first problem of processing flow generation, selecting a processing flow that corresponds to a second problem with second metadata with a certain similarity level with respect to the first metadata by referring to a memory that stores a plurality pieces of metadata and a plurality of processing flows that correspond to each of a plurality of problems, the plurality of problems including the first problem and the second problem; and specifying a tree structure of the process flow that corresponds to the first problem based on genetic programming that uses the selected processing flow as an initial individual.