CN108549607A - Message-passing parallel program Multiple path coverage test data coevolution generation method - Google Patents
Message-passing parallel program Multiple path coverage test data coevolution generation method Download PDFInfo
- Publication number
- CN108549607A CN108549607A CN201810343192.4A CN201810343192A CN108549607A CN 108549607 A CN108549607 A CN 108549607A CN 201810343192 A CN201810343192 A CN 201810343192A CN 108549607 A CN108549607 A CN 108549607A
- Authority
- CN
- China
- Prior art keywords
- population
- individual
- migration
- test data
- path
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/3668—Software testing
- G06F11/3672—Test management
- G06F11/3688—Test management for test execution, e.g. scheduling of test suites
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/3668—Software testing
- G06F11/3672—Test management
- G06F11/3676—Test management for coverage analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Hardware Design (AREA)
- Genetics & Genomics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Physiology (AREA)
- Debugging And Monitoring (AREA)
Abstract
The invention discloses a kind of message-passing parallel program Multiple path coverage test data coevolution generation methods, it is intended to the test data of covering multi-goal path is efficiently produced for message-passing parallel program.It is as follows:(1) it is that each destination path under program each schedule sequences build corresponding population respectively, the individual in population be that the program after encoding inputs;(2) design population performance and individual Performance Evaluating Indexes;(3) each Evolution of Population is solved using genetic algorithm, makes individual tend to be migrated to the good population of performance by individual migration in this course;(4) evolution for corresponding to population according to coverage goal path is stopped per generation evolution result, until generating the test data of all destination paths of covering or reaching maximum evolutionary generation, termination algorithm.
Description
Technical field
This patent belongs to software test field, and in particular to a kind of message-passing parallel program Multiple path coverage test data
Coevolution generation method can be used for generating the test data for covering a plurality of destination path in software test.
Background technology
Software test is an important component of field of software engineering, selects an outstanding test method that can have
Effect ensures software quality.In numerous method for testing software, a kind of common method is to generate the test data of high quality, and make
Program is executed with test data, finds defect or mistake present in program operation process in this way.But it uses
There are two difficulties for such method:First, it is more to generate the time that effective test data is spent;Second is that the matter of test data
Amount is difficult to be guaranteed.If suitable method can be taken, with the test data of smaller cost creation high quality, then,
Testing efficiency can be significantly improved, and reduces the expense of software test.
Concurrent program, the i.e. program containing two or more parallel executive process, due to simple development scheme, reality
Existing convenience and outstanding compatibility performance, are increasingly paid attention to by program developer, have been successfully applied to include figure
As various fields such as processing, ocean simulation, chemical processes.The test problem of concurrent program is also gradually taken seriously.And stroke
Sequence and the important difference of serial program are that have numerous schedule sequences, i.e., the dispatching sequence of process is formed in program process
Sequence.The presence of schedule sequences so that the execution of concurrent program has uncertainty, i.e., the scheduling sequence that identical program input executes
Row are different, it is possible to different operation results are obtained, so that the test of concurrent program is increasingly complex.
In a variety of parallel Programming methods, it is the most commonly used that traditional programming language is extended using Message Passing Environment,
Message passing interface therein is also current widest Parallel Programming Environment, has become the international standard of concurrent program.Cause
This, the very worth research of Test data generation problem of this message-passing parallel program.
The path coverage test data generation problems of this class method can be solved by genetic algorithm, and common method is
Using similarity of paths as value function is adapted to, by static analysis to schedule sequences yojan, and the best scheduling of performance is chosen
Sequence is built a population according to the schedule sequences, then is evolved using genetic algorithm and generate test data.If there are a plurality of targets
When path, common method is to sequentially generate the test data of every destination path of covering, or select suitable individual fitness
Calculation is the test data for producing a plurality of destination path of covering it is expected to execute an algorithm.However, these methods do not have
There is the information for making full use of concurrent program schedule sequences to provide, does not account for the difference of performance between schedule sequences, algorithm is caused to be imitated
Rate is less than satisfactory.
Coevolution is a kind of genetic algorithm evolved using multiple populations, can efficiently use information on multiple populations.
During evolution, by evaluating individual and the adaptability of population and the performance of different population, make individual to the good kind of performance
Group's migration, can improve the directionality of evolution.For parallel program testing data generation problems, if in this way, it will
The schedule sequences of concurrent program correspond to population, convert program input to the individual in population, and evaluate during evolution
The performance of schedule sequences makes program input gradually to the corresponding population migration of the good schedule sequences of performance, then, it can be rationally sharp
With schedule sequences information, the efficiency of Test data generation is improved.
Invention content
The present invention is built pair respectively for each destination path under message-passing parallel program each schedule sequences
Answer population;Then, the method for evaluating performance of population and individual is provided;Then, according to population and individual performance, genetic algorithm into
Reasonable enforcement individual migration operates during change;Finally, it evolves and generates the test data of each destination path of covering.
The technical problems to be solved by the invention:Overcome the shortcomings of existing method, a kind of test data generating method be provided,
To improve the efficiency that message-passing parallel program generates Multiple path coverage test data, testing cost is reduced.
Technical scheme of the present invention:Propose a kind of message-passing parallel program Multiple path coverage test data coevolution
Generation method, it is characterised in that following steps:
Step 1:Build population
Correspondence is built respectively for each destination path under tested message-passing parallel program each schedule sequences
Population carries out positive integer serial number to population, and parameter j used below and k indicates kind of a group number;Individual in population is to compile
Program input after code carries out positive integer serial number independently to the individual in each population, and parameter i used below is indicated
Individual number.
Step 2:The performance evaluation of population and individual
It needs to instruct evolutionary process by adapting to value function when generating test data using genetic algorithm, selects path
Similarity is as adaptation value function, and the adaptive value of individual i is denoted as F in the population j being calculatedj,i, Fj,i∈[0,1].In addition, this
Invention also needs to calculate individual relative adaptation value and Population adaptation value to instruct individual migration.For i-th of individual of population j, note
The minimum and maximum value of individual individual fitness is F in the populationj,minWith Fj,max, Fj,min,Fj,max∈ [0,1], then, kind
The individual relative adaptation value of individual i, is denoted as F ' in group jj,i, calculation formula is:
Wherein, F 'j,i∈[0,1].The value is bigger, then relative performance of the individual in affiliated population is better.
Consider that all individuals in population, the average value of these individual fitnesses can reflect population to a certain extent
Performance, therefore, as evaluation population performance an index.For population j, if the population includes mjIndividual,
So, the adaptive value of the population, is denoted as Fj, Fj∈ [0,1], calculation formula are:
F′j,iWith FjThe complexity of individual destination path test data corresponding with population generation covering is reflected respectively,
Value is bigger, then easier generation test data, performance are better.
It is to be operated to execute individual migration, and individual migration is grasped due to calculating individual relative adaptation value and Population adaptation value
Make population often evolution λ for when execute primary, therefore, the calculating of individual relative adaptation value and Population adaptation value also only need to be in population
Often evolution λ for when carry out it is primary, wherein λ is positive integer parameter.
Step 3:Individual migration operates
Individual migration refers to that the individual in a population is transferred in another population and is subsequently evolved, and individual is former
Population belonging to elder generation is known as source population, and the population that individual will be transferred to is known as target population.It is a for Multiple path coverage problem
Body migration can be divided into two kinds:Schedule sequences are migrated and destination path is migrated, individual migration is respectively referred to and extremely indicates the not people having the same aspiration and interest
The population in degree series same target path and individual migration are to the population for indicating identical schedule sequences different target path.In order to
Migration risk is reduced, in population every evolution λ generations, individual migration is judged whether to all individuals, in addition, setting population poor performance
Different degree Δj,kIndicate the diversity factor of population j and population k, Δj,k∈ [0,1], calculation formula are:
Regulation works as Δj,kWhen >=β, for population j and population k there are significant difference, wherein β is the parameter more than 0.In implementation pair
When schedule sequences migrate, only there are just migrated when significant difference for population performance;When implementation migrates destination path, only
Individual relative performance is just migrated when poor.
The individual relative performance parameter alpha of note1,α2, migration probability parameter θ, α1,α2, θ ∈ [0,1], for i-th of population j
The individual migration strategy of individual, proposition is as follows:
If F 'j,i< α1, then, one kind that identical scheduling serial different target path is indicated with population j of random selection
Group is target population, and individual carries out individual migration with probability θ;
If individual does not migrate, and F 'j,i< α2, then, the serial phase of the scheduling different from population j expressions of random selection one
Population with destination path is target population and compares the two Population adaptation value, if the Population adaptation value of target population is higher than source
Population and there are significant differences, then carry out individual migration, otherwise without individual migration;
If individual does not migrate yet, this individual is without migration.
Step 4:Generate test data
Due to not necessarily being generated in the same generation to different target coordinates measurement test data, ought be successfully generated
After the test data for covering some destination path, the population of the corresponding destination path should stop evolving, in order to avoid waste test money
Source.In next evolutionary process, the population for stopping evolving is no longer participate in follow-up evolution.
When the test data for covering all destination paths has generated or has reached the maximum evolutionary generation of setting, algorithm
It terminates, exports result.
Compared with prior art, the beneficial effects of the invention are as follows:
1, concurrent program information is rationally utilized, generates test data in the good schedule sequences of performance as much as possible, is improved
Test data generation efficiency;
2, schedule sequences performance is voluntarily judged by algorithm without manual analysis, is carried out convenient for the automation of test;
3, it is the test data for producing a plurality of destination path of covering to execute an algorithm, improves Test data generation
Effect.
Description of the drawings
Fig. 1 is the general flow chart of the present invention;
Fig. 2 is the program code for defining individual collections;
Fig. 3 is the program code for defining population set;
Fig. 4 is the program code for realizing individual migration operation;
Specific implementation mode
The part makes to show a C language program, realizes method proposed by the present invention, and combine specific attached drawing and example pair
It is described in detail.
Fig. 1 is a kind of message-passing parallel program Multiple path coverage test data coevolution generation side proposed by the present invention
The flow chart of method, this method include:
Step 1:Build population
For the message-passing parallel program containing y destination path of x schedule sequences, xy population is built, respectively
Indicate that each destination path under each schedule sequences, the individual in population are the program input after coding.For all
Individual builds individual collections respectively, and the element in set is the specifying information of corresponding individual;Kind is built respectively for all populations
Cluster is closed, and the element in set is the specifying information of corresponding population.
In C language, individual collections individual is defined using structure struct, as shown in Fig. 2, each in structure
Variable meaning such as following table:
Population set pop is defined using structure struct, as shown in figure 3, each variable meaning such as following table in structure:
Wherein, variable px, siz0 and mx assignment in initialization of population, no longer changes later.Particularly, at the beginning of variable stx
Initial value is 0, indicates that population is being evolved, and when its value is 1, the population stops evolving.
Step 2:The performance evaluation of population and individual
After being evolved per λ using genetic algorithm to test data, all individuals of traversal individual collections individual lead to
Variable fitness is crossed to calculate with gro and update the value of variable minfit, maxfit and avf in corresponding population set pop;It connects
All individuals for traversal individual collections individual again, are calculated using individual relative adaptation value calculation formula and update change
Measure the value of fij.These variables all after the completion of update, that is, complete the performance evaluation of a population and individual.
Step 3:Individual migration operates
After being evolved per λ using genetic algorithm to test data, individual migration operation is executed, is traversed in individual collections
Each element, respectively execute individual migration operation.The C programmer code for executing individual migration operation is as shown in Figure 4.Its
In, global variable f indicates that unlapped destination path quantity, constant MQ indicate that schedule sequences number, constant MN indicate destination path
Number, constant SIZE indicate that individual sum, constant A1 indicate α1, constant AX expressions θ, constant A2 indicate α2, constant B1 expressions β.
After the completion of function call, all individuals of traversal individual collections individual, the value according to variable gro calculates
And the value of the variable pno and siz in corresponding population set pop are updated, all an individual migration is completed after update
Operation.
Step 4:Generate test data
After being successfully generated the test data for covering some destination path, in population set pop, by all correspondences
The stx values of the population of the destination path are set as 1, and subtract 1 by the value of global variable f, then continue to evolve, all until covering
When the test data of destination path has generated or reached the maximum evolutionary generation of setting, algorithm terminates, and exports the survey of generation
Try data.
Next by taking a message-passing parallel program as an example, illustrate the validity of institute's extracting method of the present invention.
Selected program includes 6 inputs, 5 processes, 14 communication statements, 6 schedule sequences, input range [0,64].
2,3,4,5 destination paths are selected to carry out Test data generation respectively.Genetic algorithm uses binary coding, single-point to intersect, and hands over
Pitch probability 0.8, single-point variation, mutation probability 0.4, roulette selection, maximum 2000 generations of evolution.According to schedule sequences and target road
Diameter generates the population of corresponding number, each population scale is 20.Remaining parameter setting is as shown in the table.
Control methods is selected without individual migration, other steps method identical with institute's extracting method of the present invention.Identical
10 institute's extracting methods of the present invention are executed under environment respectively and calculate the success rate and mean time for generating test data with control methods
Between.Wherein, success rate refers to that the quantity in test data success coverage goal path accounts for the percentage of general objective number of paths, average
Time refers to generating one group of average time successfully covered needed for all destination path test datas.Experimental result such as following table institute
Show.
As seen from table, institute's extracting method of the present invention can efficiently produce the survey of covering message-passing parallel program multi-goal path
Try data.
Claims (1)
1. message-passing parallel program Multiple path coverage test data coevolution generation method, which is characterized in that this method packet
Include following steps:
Step 1:It is built respectively pair for each destination path under tested message-passing parallel program each schedule sequences
Population is answered, positive integer serial number is carried out to population, parameter j used below and k indicates that kind of a group number, the individual in population are
Program input after coding carries out positive integer serial number independently, parameter i tables used below to the individual in each population
Show individual number;
Step 2:It is independently evolved each population using genetic algorithm, selects similarity of paths as value function is adapted to, be calculated
The adaptive value of individual i is denoted as F in population jj,i, Fj,i∈[0,1];Define F 'j,iFor the individual relative adaptation of individual i in population j
Value, F 'j,i∈ [0,1], calculation formula is as follows:
In formula, Fj,minWith Fj,maxThe minimum value and maximum value of the existing all individual fitnesses of respectively population j, Fj,min,Fj,max∈
[0,1];
Define FjFor the Population adaptation value of population j, Fj∈ [0,1], calculation formula is as follows:
In formula, population scale mjTo include the quantity of individual in population j;
Genetic algorithm evolution λ generations are often used, the individual relative adaptation value F ' of all individuals is calculated and recordj,i, calculate and record institute
There is the Population adaptation value F of populationj;Wherein, λ is positive integer parameter;
Step 3:It is calculated in step 2 and has recorded the individual relative adaptation value F ' of all individualsj,iWith the Population adaptation of all populations
Value FjLater, following individual migration is judged whether to all individuals successively:
Individual migration refers to that (or multiple) individual in a population is transferred in another population and is subsequently evolved, a
Population belonging to body is original is known as source population, and the population that individual will be transferred to is known as target population;Individual migration is divided into two kinds:
Schedule sequences are migrated and destination path is migrated, respectively refer to individual migration to indicating different schedule sequences same targets path
Population and individual migration define the population property of population j and population k to the population for indicating identical schedule sequences different target path
It can diversity factor Δj,k, Δj,k∈ [0,1], calculation formula are:
F in formulakFor population k Population adaptation value, it is specified that working as Δj,kWhen > β, there are significant difference, wherein β is population j and population k
Parameter more than 0;The individual relative performance parameter alpha of note1,α2, migration probability parameter θ, α1,α2, θ ∈ [0,1], for of population j
Body i, individual migration strategy are as follows:
If F 'j,i< α1, then, random selection one indicates that the population in identical scheduling serial different target path is with population j
Target population, individual carry out individual migration with probability θ;
If individual does not migrate, and F 'j,i< α2, then, the serial same target of the scheduling different from population j expressions of random selection one
The population in path is target population and both compares Population adaptation value, if the Population adaptation value of target population higher than source population and
There are significant differences, then individual migration are carried out, otherwise without individual migration;
If individual does not migrate yet, this individual is without migration;
Step 4:After certain generation of evolution has been successfully generated the test data for covering some destination path, stop evolution pair
Should destination path population;It evolves when the test data for covering all destination paths has generated or reached the maximum of setting
When algebraically, algorithm terminates, and exports result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810343192.4A CN108549607B (en) | 2018-04-17 | 2018-04-17 | Message transmission parallel program multi-path coverage test data coevolution generation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810343192.4A CN108549607B (en) | 2018-04-17 | 2018-04-17 | Message transmission parallel program multi-path coverage test data coevolution generation method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108549607A true CN108549607A (en) | 2018-09-18 |
CN108549607B CN108549607B (en) | 2021-05-18 |
Family
ID=63515347
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810343192.4A Active CN108549607B (en) | 2018-04-17 | 2018-04-17 | Message transmission parallel program multi-path coverage test data coevolution generation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108549607B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110083531A (en) * | 2019-04-12 | 2019-08-02 | 江西财经大学 | It improves the shared multi-goal path coverage test method of individual information and realizes system |
CN110135584A (en) * | 2019-03-30 | 2019-08-16 | 华南理工大学 | Extensive Symbolic Regression method and system based on self-adaptive parallel genetic algorithm |
CN112817841A (en) * | 2021-01-21 | 2021-05-18 | 西安交通大学 | Method for determining MPI parallel program test data generation based on communication of path coverage |
CN113742251A (en) * | 2021-11-08 | 2021-12-03 | 山东建筑大学 | Software test path generation method and system based on set evolution |
CN116089317A (en) * | 2023-04-10 | 2023-05-09 | 江西财经大学 | Multipath testing method and system based on path similarity table and individual migration |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103902457A (en) * | 2014-04-17 | 2014-07-02 | 中国矿业大学 | Method for generating test data covering parallel program paths based on coevolution |
CN105718368A (en) * | 2016-01-15 | 2016-06-29 | 西安邮电大学 | Software testing data amplification method |
CN106095669A (en) * | 2016-06-03 | 2016-11-09 | 中国矿业大学 | Parallel program testing method based on schedule sequences yojan |
CA2945458A1 (en) * | 2015-11-10 | 2017-05-10 | Accenture Global Solutions Limited | Constraint extraction from natural language text for test data generation |
-
2018
- 2018-04-17 CN CN201810343192.4A patent/CN108549607B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103902457A (en) * | 2014-04-17 | 2014-07-02 | 中国矿业大学 | Method for generating test data covering parallel program paths based on coevolution |
CA2945458A1 (en) * | 2015-11-10 | 2017-05-10 | Accenture Global Solutions Limited | Constraint extraction from natural language text for test data generation |
CN105718368A (en) * | 2016-01-15 | 2016-06-29 | 西安邮电大学 | Software testing data amplification method |
CN106095669A (en) * | 2016-06-03 | 2016-11-09 | 中国矿业大学 | Parallel program testing method based on schedule sequences yojan |
Non-Patent Citations (1)
Title |
---|
巩敦卫: "基于协同进化遗传算法的消息传递并行程序路径覆盖测试数据生成", 《第七届中国测试学术会议论文集》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110135584A (en) * | 2019-03-30 | 2019-08-16 | 华南理工大学 | Extensive Symbolic Regression method and system based on self-adaptive parallel genetic algorithm |
CN110135584B (en) * | 2019-03-30 | 2022-11-18 | 华南理工大学 | Large-scale symbolic regression method and system based on adaptive parallel genetic algorithm |
CN110083531A (en) * | 2019-04-12 | 2019-08-02 | 江西财经大学 | It improves the shared multi-goal path coverage test method of individual information and realizes system |
CN110083531B (en) * | 2019-04-12 | 2022-12-27 | 江西财经大学 | Multi-target path coverage test method and implementation system for improving individual information sharing |
CN112817841A (en) * | 2021-01-21 | 2021-05-18 | 西安交通大学 | Method for determining MPI parallel program test data generation based on communication of path coverage |
CN112817841B (en) * | 2021-01-21 | 2022-12-09 | 西安交通大学 | Method for determining MPI parallel program test data generation based on communication of path coverage |
CN113742251A (en) * | 2021-11-08 | 2021-12-03 | 山东建筑大学 | Software test path generation method and system based on set evolution |
CN116089317A (en) * | 2023-04-10 | 2023-05-09 | 江西财经大学 | Multipath testing method and system based on path similarity table and individual migration |
Also Published As
Publication number | Publication date |
---|---|
CN108549607B (en) | 2021-05-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108549607A (en) | Message-passing parallel program Multiple path coverage test data coevolution generation method | |
CN112035116B (en) | Agent modeling method for multi-target compiling optimization sequence selection | |
CN114186749B (en) | Flexible workshop scheduling method and model based on reinforcement learning and genetic algorithm | |
CN105929690B (en) | A kind of Flexible Workshop Robust Scheduling method based on decomposition multi-objective Evolutionary Algorithm | |
Prakash et al. | Modified immune algorithm for job selection and operation allocation problem in flexible manufacturing systems | |
CN103235743B (en) | A kind of based on decomposing and the multiple goal test assignment dispatching method of optimum solution follow-up strategy | |
CN104239213A (en) | Two-stage scheduling method of parallel test tasks facing spacecraft automation test | |
CN101901425A (en) | Flexible job shop scheduling method based on multi-species coevolution | |
CN102331966A (en) | Software test data evolution generation system facing path | |
CN106796533A (en) | It is adaptive selected the system and method for execution pattern | |
CN111832101A (en) | Construction method of cement strength prediction model and cement strength prediction method | |
CN110135584A (en) | Extensive Symbolic Regression method and system based on self-adaptive parallel genetic algorithm | |
CN108985617A (en) | A kind of product manufacturing process dispatching method and system based on intelligence manufacture | |
Liu et al. | A sphere-dominance based preference immune-inspired algorithm for dynamic multi-objective optimization | |
CN110909787A (en) | Method and system for multi-objective batch scheduling optimization based on clustering evolutionary algorithm | |
CN114360662A (en) | Single-step inverse synthesis method and system based on two-way multi-branch CNN | |
CN109271320A (en) | A kind of upper multiple target priorities of test cases sort method | |
US20070179917A1 (en) | Intelligent design optimization method and system | |
Zhou et al. | Automatic design of scheduling policies for dynamic flexible job shop scheduling by multi-objective genetic programming based hyper-heuristic | |
CN109299776A (en) | One seed nucleus incremental transfinites learning machine, difference grey wolf method for mixing and optimizing on multiple populations | |
CN106919504B (en) | Test data evolution generation method based on GA algorithm | |
CN109034479B (en) | Multi-target scheduling method and device based on differential evolution algorithm | |
CN116432570A (en) | Method and device for generating test case of chip and storage medium | |
CN110222816A (en) | Method for building up, image processing method and the device of deep learning model | |
CN108345942A (en) | A kind of machine learning recognition methods based on embedded coding study |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |