Unlock instant, AI-driven research and patent intelligence for your innovation.

Evolution multi-task scheduling optimization method based on explicit migration

An optimization method and multi-task technology, applied in the information field, can solve problems such as inoperability

Pending Publication Date: 2021-04-16
CHONGQING UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, different optimization tasks usually have unique properties and need to rely on evolutionary mechanisms with specific search preferences to solve problems efficiently
Therefore, the method of implicit migration has received many restrictions on the selection of optimizers for different tasks, and it is not feasible to optimize multi-tasks in some complex situations.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Evolution multi-task scheduling optimization method based on explicit migration
  • Evolution multi-task scheduling optimization method based on explicit migration
  • Evolution multi-task scheduling optimization method based on explicit migration

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0076]Evolutionary single target multi-task scheduling optimization method based on explicit migration, including the following steps:

[0077]1) Determine the task OP1, Task OP2, Task OP1China Target Function and Task OP2The target function. Task OP1And task OP2The number of task targets is 1.

[0078]2) Set the infrastructure of the evolution algorithm. The infrastructure includes an evolution optimizer for a task.

[0079]The base parameters also include optimizing termination conditions, migration conditions, and constructing search space mapping required to decompose N, migrating solution amount S.

[0080]The migration condition reaches G in adjacent migration intervals.

[0081]3) Use noise reduction automatic encoder to learn to get task OP1And task OP2Ordered mapping between the corresponding search spaces.

[0082]Use noise reduction automatic encoder to learn to get task OP1And task OP2The steps of the ordered mapping between the corresponding search space are as follows:

[0083]3.1) Task OP...

Embodiment 2

[0108]Evolutionary Multi-Object Scheduling Optimization Method Based on Explicit Migration, including the following steps:

[0109]1) Determine the task OP1, Task OP2, Task OP1All target functions and tasks OP2All target functions. The task OP1And task OP2The number of targets is greater than 1.

[0110]2) Set the basic parameters of the multi-objective evolution algorithm. The infrastructure includes an evolution optimizer for all tasks.

[0111]The basic parameters include optimizing termination conditions, migration conditions, and constructing a solution amount N required for searching space mapping, the migration amount S.

[0112]The migration condition reaches G in adjacent migration intervals.

[0113]3) Use noise reduction automatic encoder to learn to get task OP1And task OP2Ordered mapping between the corresponding search spaces.

[0114]Use noise reduction automatic encoder to learn to get task OP1And task OP2The steps of the ordered mapping between the corresponding search spaces are as ...

Embodiment 3

[0140]A multi-task algorithm based on display migration method, characterized in that the following steps are included:

[0141]1) Initializing the basic parameters of the evolution multitasking algorithm, including setting optimization termination conditions; set the migration condition; set the solution required to build the search space map (N); set the amount of decompensation of the migration (s); set independent for all tasks Optimizer.

[0142]2) Learn the ordered mapping between the search space corresponding to any two tasks, the steps include:

[0143]Use the Denoising AutoEncoder to learn the feature spatial mapping relationship between any two tasks. For any two tasks (source task OP1, As well as the target task OP2), P and Q represent from OP1OP2A set of solutions for uniform sampling in the space. Where P = {P1, P2... pN}, And q = {q1, Q2...QN}, N is the size of the decoction. From OP1Go to OP2The mapping relational M can be built by a noise reduction automatic encoder, and the...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses an evolution multitask scheduling optimization method based on explicit migration, and the method comprises the steps: 1) determining all multitasks needing to be optimized and all target functions in each task; 2) setting basic parameters of an evolutionary algorithm; 3) performing learning by using a noise reduction automatic encoder to obtain ordered mapping between search spaces corresponding to any two tasks; 4) carrying out independent evolution optimization on the plurality of tasks by utilizing an evolution optimizer; 5) when a migration condition is met, carrying out explicit migration between any two tasks; 6) repeating the step 4) and the step 5) until an optimization termination condition is met; and 7) outputting final optimal solutions of all tasks. According to the method, the problem that a traditional evolutionary algorithm can only solve a single task is solved, the limitation of an existing multi-task algorithm on an evolutionary optimizer is broken through, and the multi-task solving efficiency is improved.

Description

Technical field[0001]The present invention relates to the field of information technology, and in particular, an evolutionary multi-task scheduling optimization method based on explicit migration.Background technique[0002]Evolutionary Algorithm, EA is a kind of high robust adaptive search algorithm, which inspires biological evolution, and the general steps include cross, variation, option, etc. These steps are repeatedly iterated to terminate when meeting predetermined conditions. Due to powerful search capabilities and ease of use, in the past few decades, the evolutionary algorithm has made significant breakthroughs and success on solving complex optimization problems in the real world. Common evolution optimers include genetic algorithm, ga, differential evolution, de), etc., main steps include population initialization, cross, variation, and environmental choice. Population initialization often uses a uniform sample method or a latin ultrasound method (Latin Hypercube) randomly...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06F9/48
Inventor 冯亮黄灵煜周磊
Owner CHONGQING UNIV