CN107257307A - A kind of parallelization genetic algorithm for solving multiple terminals collaboration network access method based on Spark - Google Patents
A kind of parallelization genetic algorithm for solving multiple terminals collaboration network access method based on Spark Download PDFInfo
- Publication number
- CN107257307A CN107257307A CN201710514864.9A CN201710514864A CN107257307A CN 107257307 A CN107257307 A CN 107257307A CN 201710514864 A CN201710514864 A CN 201710514864A CN 107257307 A CN107257307 A CN 107257307A
- Authority
- CN
- China
- Prior art keywords
- population
- spark
- network
- terminal
- rdd
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L12/00—Data switching networks
- H04L12/28—Data switching networks characterised by path configuration, e.g. LAN [Local Area Networks] or WAN [Wide Area Networks]
- H04L12/2854—Wide area networks, e.g. public data networks
- H04L12/2856—Access arrangements, e.g. Internet access
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
Abstract
A kind of parallelization genetic algorithm for solving multiple terminals collaboration network access method based on Spark, it is adaptable to which communication terminal collection cooperates with large server.Step is：Large server confirms the quantity and network terminal parameter of access communications terminal；All network terminal parameter informations are stored in the HDFS catalogues of cluster by large server；Cutting is stored in the network terminal parameter information in HDFS catalogues in the way of Spark clusters；Parallel time genetic algorithm based on Spark clusters is carried out to the information after cutting, obtains being adapted to the final for population data of current multiple terminals collaboration access network scheme；To finally for population data optimizing, so as to obtain optimal network scheme；Parallelizing network communication is carried out using optimal network scheme and multiple communication terminals.It can realize complete parallel time genetic algorithm preferred terminal collection collaboration access network on multiple nodes of Spark clusters, reduce the complexity and amount of calculation of network terminal selection, significantly improve the selection quality and transmission rate of the network terminal.
Description
Technical field
Network access method is cooperateed with the present invention relates to a kind of multiple terminals, be particularly suitable for use in a kind of parallelization based on Spark
Genetic algorithm for solving multiple terminals cooperates with network access method
Technical background
Spark is a distributed computing framework increased income, and it is based on elasticity distribution formula data set (RDD), and RDD is using first
The directed acyclic graph mechanism entered supports looping traffic operation, and successive ignition can just be completed by importing internal memory by an iteration,
The algorithm of a large amount of iteration is so needed to be highly suitable for computing on Spark platforms similar to Genetic Algorithm.
Genetic algorithm (Genetic Algorithm) is natural selection and the science of heredity machine for simulating Darwinian evolutionism
The computation model of the biological evolution process of reason, is a kind of method by simulating natural evolution process searches optimal solution.Traditional
The operation such as intersection, variation, selection that genetic algorithm is carried out all is carried out in units of group of individuals i.e. population, for every
Individual all has certain concurrency, so genetic algorithm has natural concurrency, especially when processing large-scale dataset
When, complete parallelization processing can be achieved between each sub- population.
Multiple terminals contract network refers to share wan communication ability in the way of collaboration between multiple terminals, common for use
Family provides business, so as to break away from limitation of the single terminal in terms of performance, disposal ability, realizes terminal and Internet resources
Efficiently utilize.
With continuing to develop for communication network technology and terminal device, the selectable terminal of customer service and access way by
It is gradually diversified.In order to overcome the performance bottleneck of single wireless access technology, the terminal and Internet resources around user are made full use of,
Multiple terminals communication for coordination pattern is arisen at the historic moment.Under the multi-network environment of multiple terminals, how to select suitable terminal is user service
It is a problem, and the selection of the genetic algorithm parallelization multi-terminal network based on Spark can be effectively improved.
The content of the invention
For above-mentioned technical problem there is provided a kind of method is simple, execution efficiency is high, effectively reduces the network terminal preferred
Amount of calculation and complexity, realize the parallelization genetic algorithm for solving based on Spark of parallel time genetic algorithm on multiple nodes
Multiple terminals cooperates with network access method.
To realize above-mentioned technical purpose, the parallelization genetic algorithm for solving multiple terminals collaboration of the invention based on Spark connects
Enter network method, the suitable multiple communication terminal collection of selection cooperate with large server performs data service, and step is as follows：
Large server confirms to access the quantity of each communication terminal and the network terminal parameter of each communication terminal；
All network terminal parameter informations are stored in the HDFS catalogues of cluster by large server；
Large server cutting in the way of Spark clusters is stored in the network terminal parameter information in HDFS catalogues；
Parallel time genetic algorithm based on Spark is carried out to the network terminal parameter information after cutting, so as to be adapted to
Current multiple terminals cooperates with the final for population data of access network scheme；
To finally for population data optimizing, so as to obtain being best suitable for current multiple terminals collaboration access network scheme；
Large server is whole using the multiple communications being best suitable in current multiple terminals collaboration access network scheme and scheme
End carries out parallelizing network communication.
Further selection step is as follows：
The all-network terminal parameter of large server is carried out genetic algorithm initialization by step 1. according to demand, will be initial
Variable parameter after change is converted to binary system, and the parametric variable after each binarization is stored in greatly line by line as individual specimen
The initial population sample data of all communication terminals is formed in the HDFS of type server cluster；
Step 2. seeks the quantity set Spark clustered node quantity of terminal as needed, by initial population sample data
Multiple sons based on Spark clusters are averagely cut into the way of the individual sum/Spark clustered node numbers of population sample data to plant
Group；
Step 3. is carried out using genetic algorithm to each terminal parameter variable in multiple sub- populations based on Spark clusters
Global evolution optimizing, upsets filial generation each sub- population at individual distribution in phylogenetic scale in the method reordered, merges all sub- kinds
Natural selection is carried out after group's individual, obtains final for population at individual；
Step 4. is carried out using Spark APIs functions reduceByKey to all final fitness for population at individual
Sequence, obtains multiple optimum individuals of demanding terminal needed for global fitness meets, then obtain end to the decoding of multiple optimum individuals
Weighting parameter is held, optimized parameter collection is selected from candidate terminal according to optimum terminal weighting parameter value, compares present terminal collection institute
The data rate R that can be providedcWhether user's request data rate R is mett, if meet if terminate selection, otherwise return to step 1 after
It is continuous to choose, cooperate with access network scheme until choosing optimal termination set as optimal multiple terminals；
Step 5. large server cooperates with access network scheme to be carried out simultaneously with corresponding communication terminal using optimal multiple terminals
Rowization network service.
The method reordered described in step 3 upsets filial generation in phylogenetic scale and each merges all sons after sub- population at individual is distributed
The step of population carries out natural selection can carry out successive ignition circulate operation as needed, and the step makes individual in evolution space
Disperse as far as possible, keep the independence of parameter and population individual, it is to avoid a large amount of individuals all point to same terminal parameter in population, so that
The accuracy rate that terminal parameter is finally taken for individual species mass selection is effectively improved, final generation is exported after default iterations is met
Body population carries out subsequent operation.
Carry out global evolution optimizing and obtain final concretely comprising the following steps for population at individual:
Step 3.1：Start APIs functions map generation evolutionary process, each cluster in the sub- population based on Spark clusters
Node creates population RDD to the sub- population after cutting respectively；
Step 3.2：Crossover operation is performed to population RDD using function map, new population RDD ' is generated；
Step 3.3：Population RDD ' the execution mutation operation new to generating obtains population RDD "；
Step 3.4：Calculate the relative adaptation angle value of each individuals of newly-generated population RDD " respectively according to fitness function,
By the fitness of individual with parametric variable with key-value pair<Fitness, parameter>Form stored；
Step 3.5：Merge all sub- populations, according to the relative adaptation angle value of individual, perform roulette algorithm operating, choosing
Go out to enter follow-on optimum individual；
Step 3.6：If meeting default iterations condition, the population in final generation is exported, is terminated；If being unsatisfactory for iterations
Condition, performs step 3.2 and carries out next round iteration.
The step 3.2 is further：
By defining population RDD global listings, crossover operation is carried out to global sample on each Spark clustered nodes,
The whole population RDD created on each Spark clustered node are sampled using function take, it is flat using function parallelize
Store into two population RDD, intersected two population RDD as random pair by constituting the form of key-value pair
Two parents；Chromosome is judged by way of producing random number to whether intersecting, judged result is intersected with predetermined
Probability PCThe chromosome pair selected and need to carry out crossover operation in population is compared, it is then random to determine chromosome to a certain base
Because the position after seat is crosspoint, the chromosome dyads of two pairing individuals are exchanged with each other in the point, the two are new for final output
Chromosome, replaces old chromosome by new chromosome and generates new population RDD '.
The step 3.3 is further：
Mutation operation is carried out to completing the population RDD ' after crossover operation：Read one by one using function map and intersect what is produced
Chromosome coding, travels through each locus of each chromosome and produces random number, by itself and predetermined mutation probability PMEnter
Row, which compares to select, needs the locus of progress mutation operation to be judged in chromosome, the locus for meeting variation condition is carried out
Inversion operation, so that the new chromosome object of output one；Otherwise the chromosome object is directly exported, is ultimately produced after variation
Population RDD ".
Beneficial effect：The application utilizes characteristic of the Spark clusters based on internal memory distributed arithmetic, the network after initialization
Terminal parameter group by (population at individual sum/Spark clustered nodes number) be averagely cut into many individual sub- Species structures cluster each
Parallelization computing is realized in node, reorders and upsets phylogenetic scale filial generation each sub- population at individual, makes it most in evolution space
It may disperse, keep the independence of parameter and population individual, it is to avoid a large amount of individuals all point to same terminal parameter in population, occur
The problem of terminal parameter is repeated, and by having given full play to the potential of genetic algorithm, analysis is adapted to the terminal of present access network
Information parameter, therefrom fast selecting meet the optimum terminal collection of demand, complete terminal selection.This method execution efficiency is high, effectively
The network terminal preferred amount of calculation and complexity are reduced, the service quality of business when improving selection multiple terminals collaboration access network
With the high efficiency of transmission.
Brief description of the drawings
Fig. 1 is the overall flow figure of the present invention；
Fig. 2 is the crossover process schematic diagram of the present invention；
Fig. 3 is the mutation process schematic diagram of the present invention；
Fig. 4 is the selection multiple terminals collaboration access network process schematic diagram of the present invention；
Embodiment
One embodiment to the application is described further below in conjunction with the accompanying drawings：
As shown in figure 1, the collaboration access network side of the parallelization genetic algorithm for solving multiple terminals based on Spark of the present invention
Method, the suitable multiple communication terminal collection of selection cooperate with large server performs data service, and step is as follows：
Large server confirms to access the quantity of each communication terminal and the network terminal parameter of each communication terminal；
All network terminal parameter informations are stored in the HDFS catalogues of cluster by large server；
Large server cutting in the way of Spark clusters is stored in the network terminal parameter information in HDFS catalogues；
Parallel time genetic algorithm based on Spark is carried out to the network terminal parameter information after cutting, so as to be adapted to
Current multiple terminals cooperates with the final for population data of access network scheme；
To finally for population data optimizing, so as to obtain being best suitable for current multiple terminals collaboration access network scheme；
Large server is whole using the multiple communications being best suitable in current multiple terminals collaboration access network scheme and scheme
End carries out parallelizing network communication.
Further selection step is as follows：
Step 1：The all-network terminal parameter of large server is subjected to genetic algorithm initialization according to demand, will be initial
Variable parameter after change is converted to binary system, and the parametric variable after each binarization is stored in greatly line by line as individual specimen
The initial population sample data of all communication terminals is formed in the HDFS of type server cluster；
Step 2：Seek the quantity set Spark clustered node quantity of terminal as needed, by initial population sample data
Multiple sons based on Spark clusters are averagely cut into the way of the individual sum/Spark clustered node numbers of population sample data to plant
Group；
Step 3：Each terminal parameter variable in multiple sub- populations based on Spark clusters is carried out using genetic algorithm
Global evolution optimizing, upsets filial generation each sub- population at individual distribution in phylogenetic scale in the method reordered, merges all sub- kinds
Natural selection is carried out after group's individual, obtains final for population at individual, it is concretely comprised the following steps:
Step 3.1：Start APIs functions map generation evolutionary process, each cluster in the sub- population based on Spark clusters
Node creates population RDD to the sub- population after cutting respectively；
Step 3.2：Crossover operation is performed to population RDD using function map, new population RDD ' is generated：
As shown in Fig. 2 by defining population RDD global listings, being carried out on each Spark clustered nodes to global sample
The whole population RDD created on each Spark clustered node are sampled, utilize function by crossover operation using function take
Parallelize is averagely stored into two population RDD, by constitute the form of key-value pair using two population RDD as
Two parents that random pair intersects；Judge that chromosome, to whether intersecting, judges knot by way of producing random number
Fruit and predetermined crossover probability PCThe chromosome pair selected and need to carry out crossover operation in population is compared, it is then random to determine
Chromosome is crosspoint to the position after a certain locus, is exchanged with each other the chromosome dyad of two pairing individuals in the point, most
The two new chromosomes are exported eventually, and new chromosome is replaced into old chromosome generates new population RDD '；
The method reordered upsets phylogenetic scale filial generation and each merges all sub- populations progress after sub- population at individual is distributed certainly
So the step of selection can carry out successive ignition circulate operation as needed, and the step makes individual divide as far as possible in evolution space
Dissipate, keep the independence of parameter and population individual, it is to avoid a large amount of individuals all point to same terminal parameter in population, so as to effectively improve
The accuracy rate of terminal parameter is finally taken for individual species mass selection, individual population of final generation is exported after default iterations is met and is entered
Row subsequent operation.
Step 3.3：Population RDD ' the execution mutation operation new to generating obtains population RDD "：
As shown in figure 3, carrying out mutation operation to completing the population RDD after crossover operation：Friendship is read one by one using function map
The chromosome coding produced is pitched, each locus of each chromosome is traveled through and produces random number, by itself and predetermined variation
Probability PMBeing compared to select needs the locus of progress mutation operation to be judged in chromosome, the base to meeting variation condition
Because seat carries out inversion operation, so that the new chromosome object of output one；Otherwise the chromosome object is directly exported, is ultimately produced
Population RDD " after variation；
Step 3.4：Calculate the relative adaptation angle value of each individuals of newly-generated population RDD " respectively according to fitness function；
By the fitness of individual with parametric variable with key-value pair<Fitness, parameter>Form stored.
Step 3.5：Merge all sub- populations, according to the relative adaptation angle value of individual, perform roulette algorithm operating, choosing
Go out to enter follow-on optimum individual；
Step 3.6：If meeting default iterations condition, the population in final generation is exported, is terminated；If being unsatisfactory for iterations
Condition, performs step 3.2 and carries out next round iteration.
Step 4：Finally fitted as shown in Figure 4 using Spark APIs functions reduceByKey to all for population at individual
Response is ranked up, and obtains multiple optimum individuals of demanding terminal needed for global fitness meets, then to multiple optimum individual solutions
Code obtains terminal weighting parameter, and optimized parameter collection is selected from candidate terminal according to optimum terminal weighting parameter value, relatively more current
The data rate R that termination set can be providedcWhether user's request data rate R is mett, terminate selection if meeting, otherwise return
Step 1 continues to choose, and access network scheme is cooperateed with until choosing optimal termination set as optimal multiple terminals；
Step 5. large server cooperates with access network scheme to be carried out simultaneously with corresponding communication terminal using optimal multiple terminals
Rowization network service.
Claims (6)
1. a kind of parallelization genetic algorithm for solving multiple terminals collaboration network access method based on Spark, it is multiple that selection is adapted to
Communication terminal collection cooperates with large server and performs data service, it is characterised in that step is as follows：
Large server confirms to access the quantity of each communication terminal and the network terminal parameter of each communication terminal；
All network terminal parameter informations are stored in the HDFS catalogues of Spark clusters by large server；
Large server cutting in the way of Spark clusters is stored in the network terminal parameter information in HDFS catalogues；
Parallel time genetic algorithm based on Spark clusters is carried out to the network terminal parameter information after cutting, so as to be adapted to
Current multiple terminals cooperates with the final for population data of access network scheme；
To finally for population data optimizing, so as to obtain being best suitable for current multiple terminals collaboration access network scheme；
Large server is entered using current multiple terminals collaboration access network scheme is best suitable for multiple communication terminals in scheme
Row parallelizing network communicates.
2. the parallelization genetic algorithm for solving multiple terminals collaboration network access method according to claim 1 based on Spark,
It is characterized in that further selection step is as follows：
The all-network terminal parameter of large server is carried out genetic algorithm initialization by step 1. according to demand, after initialization
Variable parameter be converted to binary system, the parametric variable after each binarization is stored in large-scale clothes as individual specimen line by line
The initial population sample data of all communication terminals is formed in the HDFS of business device cluster；
Step 2. seeks the quantity set Spark clustered node quantity of terminal as needed, and initial population sample data is pressed and planted
The mode of the individual sum/Spark clustered node numbers of group's sample data is averagely cut into multiple sub- populations based on Spark clusters；
Step 3. is carried out global using genetic algorithm to each terminal parameter variable in multiple sub- populations based on Spark clusters
Evolution optimizing, upsets filial generation each sub- population at individual distribution in phylogenetic scale in the method reordered, merges all sub- populations
Natural selection is carried out after body, obtains final for population at individual；
Step 4. is ranked up using Spark APIs functions reduceByKey to all final fitness for population at individual,
Multiple optimum individuals of demanding terminal needed for global fitness meets are obtained, then terminal weights are obtained to the decoding of multiple optimum individuals
Parameter, optimized parameter collection is selected according to optimum terminal weighting parameter value from candidate terminal, is compared present terminal collection and be can be provided
Data rate RcWhether user's request data rate R is mett, terminate selection if meeting, otherwise return to step 1 continues to choose,
Access network scheme is cooperateed with until choosing optimal termination set as optimal multiple terminals；
Step 5. large server cooperates with access network scheme to carry out parallelization with corresponding communication terminal using optimal multiple terminals
Network service.
3. the parallelization genetic algorithm for solving multiple terminals collaboration network access method according to claim 2 based on Spark,
It is characterized in that：The method reordered upsets filial generation and each merges all sub- populations after the distribution of sub- population at individual in phylogenetic scale
The step of row natural selection, can carry out successive ignition circulate operation as needed, the step make individual in evolution space as far as possible
It is scattered, keep the independence of parameter and population individual, it is to avoid a large amount of individuals all point to same terminal parameter in population, so as to effectively carry
Height finally takes the accuracy rate of terminal parameter for individual species mass selection, and final generation individual population is exported after default iterations is met
Carry out subsequent operation.
4. the collaboration access network side of the parallelization genetic algorithm for solving multiple terminals based on Spark according to Claims 2 or 3
Method, it is characterised in that carry out global evolution optimizing and obtain final concretely comprising the following steps for population at individual:
Step 3.1：Start APIs functions map generation evolutionary process, each clustered node in the sub- population based on Spark clusters
Population RDD is created to the sub- population after cutting respectively；
Step 3.2：Crossover operation is performed to population RDD using function map, new population RDD ' is generated；
Step 3.3：Population RDD ' the execution mutation operation new to generating obtains population RDD "；
Step 3.4：Calculate the relative adaptation angle value of each individuals of newly-generated population RDD " respectively according to fitness function；
Step 3.5：Merge all sub- populations, according to the relative adaptation angle value of individual, perform roulette algorithm operating, select into
Enter follow-on optimum individual；
Step 3.6：If meeting default iterations condition, the population in final generation is exported, is terminated；If being unsatisfactory for iterations bar
Part, performs step 3.2 and carries out next round iteration.
5. the parallelization genetic algorithm for solving multiple terminals collaboration network access method according to claim 4 based on Spark,
It is characterized in that the step 3.2 further comprises：
By defining population RDD global listings, crossover operation is carried out to global sample on each Spark clustered nodes, utilized
Function take samples to the whole population RDD created on each Spark clustered node, is averagely deposited using function parallelize
Store up into two population RDD, by constituting two that the form of key-value pair is intersected using two population RDD as random pair
Parent；Chromosome is judged by way of producing random number to whether intersecting, judged result and predetermined crossover probability
PCThe chromosome pair selected and need to carry out crossover operation in population is compared, it is then random to determine chromosome to a certain locus
Position afterwards is crosspoint, and the chromosome dyad of two pairing individuals, the two new dyeing of final output are exchanged with each other in the point
Body, replaces old chromosome by new chromosome and generates new population RDD '.
6. the parallelization genetic algorithm for solving multiple terminals collaboration network access method according to claim 4 based on Spark,
It is characterized in that the step 3.3 further comprises：
Mutation operation is carried out to completing the population RDD ' after crossover operation：Read the dyeing for intersecting and producing one by one using function map
Body is encoded, and is traveled through each locus of each chromosome and is produced random number, by itself and predetermined mutation probability PMCompared
Relatively selecting needs to carry out mutation operation locus in chromosome is judged, the locus for meeting variation condition is negated
Operation, so that the new chromosome object of output one；Otherwise the chromosome object is directly exported, the population after variation is ultimately produced
RDD”。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710514864.9A CN107257307B (en) | 2017-06-29 | 2017-06-29 | Spark-based method for solving multi-terminal cooperative access network by parallelization genetic algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710514864.9A CN107257307B (en) | 2017-06-29 | 2017-06-29 | Spark-based method for solving multi-terminal cooperative access network by parallelization genetic algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107257307A true CN107257307A (en) | 2017-10-17 |
| CN107257307B CN107257307B (en) | 2020-06-02 |
Family
ID=60024430
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710514864.9A Active CN107257307B (en) | 2017-06-29 | 2017-06-29 | Spark-based method for solving multi-terminal cooperative access network by parallelization genetic algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107257307B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108256694A (en) * | 2018-03-02 | 2018-07-06 | 我知盘中餐（厦门）电子商务有限公司 | Based on Fuzzy time sequence forecasting system, the method and device for repeating genetic algorithm |
| CN110867224A (en) * | 2019-10-28 | 2020-03-06 | 南通大学 | Multi-granularity Spark super-trust fuzzy method for large-scale brain pathology segmentation |
| CN112785407A (en) * | 2021-02-10 | 2021-05-11 | 中国工商银行股份有限公司 | Multi-population genetic algorithm-based method and device for optimizing distribution proportion of nodes |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104866904A (en) * | 2015-06-16 | 2015-08-26 | 中电科软件信息服务有限公司 | Parallelization method of BP neural network optimized by genetic algorithm based on spark |
| US20150294237A1 (en) * | 2002-04-11 | 2015-10-15 | Gavin Stener | System and method for simultaneously displaying non-scheduled and scheduled air travel services for booking flights |
| CN105323828A (en) * | 2014-07-29 | 2016-02-10 | 中国移动通信集团河南有限公司 | Multi-mode terminal network access selection method and terminal |
| CN106502964A (en) * | 2016-12-06 | 2017-03-15 | 中国矿业大学 | A kind of extreme learning machine parallelization computational methods based on Spark |
-
2017
- 2017-06-29 CN CN201710514864.9A patent/CN107257307B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150294237A1 (en) * | 2002-04-11 | 2015-10-15 | Gavin Stener | System and method for simultaneously displaying non-scheduled and scheduled air travel services for booking flights |
| CN105323828A (en) * | 2014-07-29 | 2016-02-10 | 中国移动通信集团河南有限公司 | Multi-mode terminal network access selection method and terminal |
| CN104866904A (en) * | 2015-06-16 | 2015-08-26 | 中电科软件信息服务有限公司 | Parallelization method of BP neural network optimized by genetic algorithm based on spark |
| CN106502964A (en) * | 2016-12-06 | 2017-03-15 | 中国矿业大学 | A kind of extreme learning machine parallelization computational methods based on Spark |
Non-Patent Citations (3)
| Title |
|---|
| 李坤等: "基于Spark的LIBSVM参数优选并行化算法", 《南京大学学报（自然科学）》 * |
| 杨存: "基于多终端协同的终端与网络协同选择算法研究", 《计算机技术与发展》 * |
| 黄湖: "泛在网下多终端多网络协同的接入网络选择算法", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108256694A (en) * | 2018-03-02 | 2018-07-06 | 我知盘中餐（厦门）电子商务有限公司 | Based on Fuzzy time sequence forecasting system, the method and device for repeating genetic algorithm |
| CN110867224A (en) * | 2019-10-28 | 2020-03-06 | 南通大学 | Multi-granularity Spark super-trust fuzzy method for large-scale brain pathology segmentation |
| CN110867224B (en) * | 2019-10-28 | 2022-02-08 | 南通大学 | Multi-granularity Spark super-trust fuzzy method for large-scale brain pathology segmentation |
| CN112785407A (en) * | 2021-02-10 | 2021-05-11 | 中国工商银行股份有限公司 | Multi-population genetic algorithm-based method and device for optimizing distribution proportion of nodes |
| CN112785407B (en) * | 2021-02-10 | 2024-02-06 | 中国工商银行股份有限公司 | Lattice point distribution proportion optimizing method and device based on multiple swarm genetic algorithms |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107257307B (en) | 2020-06-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cavalcanti et al. | Combining diversity measures for ensemble pruning | |
| CN104035816B (en) | Cloud computing task scheduling method based on improved NSGA-II | |
| CN106527381B (en) | A kind of fast evaluation method towards parallel batch processing machine dynamic dispatching | |
| CN107911300B (en) | Multicast routing optimization method based on whale algorithm and application of multicast routing optimization method on Spark platform | |
| Schulz | High quality graph partitioning | |
| CN110188785A (en) | A kind of data clusters analysis method based on genetic algorithm | |
| CN102214213A (en) | Method and system for classifying data by adopting decision tree | |
| Jiang et al. | An efficient evolutionary user interest community discovery model in dynamic social networks for internet of people | |
| CN107257307A (en) | A kind of parallelization genetic algorithm for solving multiple terminals collaboration network access method based on Spark | |
| CN110647995A (en) | Rule training method, device, equipment and storage medium | |
| CN115116537B (en) | Method and system for calculating multiple transformation paths of biomolecule functional dynamics | |
| Folino et al. | GP ensembles for large-scale data classification | |
| CN106156067A (en) | For creating the method and system of data model for relation data | |
| CN110119408A (en) | Mobile object continuous-query method under geographical space real-time streaming data | |
| Wang et al. | Graph-mamba: Towards long-range graph sequence modeling with selective state spaces | |
| CN110008023B (en) | Cloud computing system budget constraint random task scheduling method based on genetic algorithm | |
| Ruan et al. | Community discovery: Simple and scalable approaches | |
| CN105005501B (en) | A kind of second order optimizing and scheduling task method towards cloud data center | |
| Nguyen et al. | A parallel hybrid genetic algorithm for multiple protein sequence alignment | |
| Czajkowski et al. | Steering the interpretability of decision trees using lasso regression-an evolutionary perspective | |
| Coelho et al. | Inducing multi-objective clustering ensembles with genetic programming | |
| Wang et al. | Lnetwork: an efficient and effective method for constructing phylogenetic networks | |
| Hu et al. | A practical design of hash functions for IPv6 using multi-objective genetic programming | |
| CN102254040A (en) | SVM (Support Vector Machine)-based Web partitioning method | |
| Newborough et al. | A generic framework for population-based algorithms, implemented on multiple FPGAs |
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 | ||
| CB02 | Change of applicant information |
Address after: 221116 Research Institute of China University of Mining and Technology,, Jiangsu Applicant after: China University of Mining & Technology Address before: 221116 Research Institute, China University of Mining and Technology, Xuzhou University, Jiangsu, China, Applicant before: China University of Mining & Technology |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |