CN101071412A - Neural network analysis system and method based on self-definition model - Google Patents
Neural network analysis system and method based on self-definition model Download PDFInfo
- Publication number
- CN101071412A CN101071412A CN 200610054283 CN200610054283A CN101071412A CN 101071412 A CN101071412 A CN 101071412A CN 200610054283 CN200610054283 CN 200610054283 CN 200610054283 A CN200610054283 A CN 200610054283A CN 101071412 A CN101071412 A CN 101071412A
- Authority
- CN
- China
- Prior art keywords
- neural network
- model
- analysis
- self
- network model
- 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.)
- Pending
Links
Images
Abstract
The invention discloses a self-defined model of artificial neural network analysis system and method. The system and method use of a powerful database system as a background, the computer automatically through artificial neural network model of arbitrary custom, and in accordance with its definition of the artificial neural network model for automatic data analysis and forecasting training. This makes the original artificial neural network professionals need to take several months of artificial neural network data analysis, only just a few minutes to complete, at the same time, as a result of a custom graphical modeling and engineering management makes artificial steps Neural network analysis more simple and fast. This invention changed the artificial neural network analysis by completely relying on the traditional method of programming, Neural network analysis makes further leap to the intelligent, personalized self-definition stage. Through the system and method for the analysis of the provision of professional staff, and rapid analysis of neural network services, thereby greatly improving the artificial neural network analysis of quality and efficiency.
Description
[technical field]
The present invention relates to a kind of artificial neural network analysis system and method for self-definition model, particularly relate to a kind of need not professional artificial neural network professional and the system and method for setting up artificial nerve network model fast of hand-coding artificial nerve network model program.
[background technology]
Because the artificial neural network analysis method has and can carry out massively parallel processing to information, has very strong robustness, non-linear and fault-tolerance, be good at association, summarize, analogy and reasoning, and have characteristics such as very strong self study and memory capability and be good at from a large amount of statistical data analyzing and extract the macroscopic statistics rule, so artificial neural network analysis is applied to teaching and scientific research in extensively, experimental analysis, Machine Design, civil engineering work, biomedical, Aero-Space, hydrometeorology, hazard prediction, the electric power experiment, automobile, electronic product, intelligence instrument, security, finance, all trades and professions such as telecommunications.But because the user generally is not the artificial neural network professional of specialty, and often requiring to set up multiple different artificial nerve network model, the user carries out the analyses and prediction of data, because the corresponding artificial nerve network model of nobody's artificial neural networks professional's Custom Design makes and much researchs and analyses all and can't carry out.
Along with network technology and development of computer, network can become the platform of artificial neural network analysis operation, is indicating that the artificial neural network technology means are by relying on artificial programming further to cross intellectuality, personalized self-defining stage fully.In order to solve the problem that exists in the application of artificial neural network analysis method, the present invention proposes a kind of artificial neural network analysis system and method for self-definition model.
[summary of the invention]
The objective of the invention is to finish automatically any customization of artificial nerve network model, and realize automaticdata training and forecast analysis according to its self-defining artificial nerve network model by computing machine.This makes needed artificial neural network professional artificial neural network data analysis work consuming time several months originally, only needed to finish in short several minutes, simultaneously, owing to adopted self-defined figure modeling and through engineering approaches management process to make artificial neural network analysis simple more, quick.
The invention provides a kind of artificial neural network analysis system and method for self-definition model, a plurality of client computers can be set up the project that will analyze according to the through engineering approaches management process, set up self-defining artificial nerve network model fast by dragging the node figure of representing neural network model then.In the artificial neural network project that defines, can be associated with on corresponding training sample data and the analysis data according to self-defining artificial nerve network model, according to different artificial nerve network model types, select different neural network algorithms to carry out computing, thereby realize quick training and data analysis prediction.A plurality of client computers can be inquired about the type and the parameter of neural network model, editor and deletion action, also can carry out association fast, realize artificial neural network modeling analysis flow process intelligent, robotization, reduce the difficulty of using analysis of neural network data message.Adopt three-tier architecture based on the artificial neural network analysis system and method for self-definition model, i.e. database server, application server and a plurality of client computer are set up the system of data centralized stores and distribution applications.
Utilize the artificial neural network analysis system and method for self-definition model provided by the invention, can tackle analyze demands complicated in the real world applications, can also select suitable algorithm and then reach best forecast analysis effect according to analysis field, industry characteristic and concrete condition.
[description of drawings]
Fig. 1 is the system architecture diagram of the artificial neural network analysis system and method for a kind of self-definition model of the present invention.
Fig. 2 is the application server functionality module map of the artificial neural network analysis system and method for a kind of self-definition model of the present invention.
Fig. 3 is the self-defined artificial nerve network model hoist pennants of the artificial neural network analysis system and method for a kind of self-definition model of the present invention.
Fig. 4 is the analysis process figure of the artificial neural network analysis system and method for a kind of self-definition model of the present invention.
[embodiment]
As shown in Figure 1, be the system architecture diagram of the artificial neural network analysis system and method for a kind of self-definition model of the present invention.The artificial neural network analysis system 6 of this self-definition model comprises a database server 1, an application server 2, network 3 and many client computers 4, wherein many client computers 4 are linked to each other with application server 2 and database server 1 by network 3, realize the information interchange of self-defined foundation of neural network model and data training analysis.The artificial neural network analysis system 6 of this self-definition model by the through engineering approaches management process based on project, the artificial nerve network model information stores of the different definition model under the project in database server 1, its objective is to guarantee that neural network model and project analysis process combine closely, make up a quick analysis system that meets various concrete artificial neural network analysis models. Application server 2 is used for receiving the neural network model information and the parameter of many client computers 4, integrates the information that generates self-defined neural network model table 5; Many client computers 4, realize the foundation of graphical nodes neural network model, the inquiry of model parameter, setting, modification and deletion or the like and data training and the analysis of carrying out neural network, and can output to many client computers 4 to analysis result by application server 2.
As shown in Figure 2, be the application server functionality module map of the artificial neural network analysis system and method for a kind of self-definition model of the present invention.These application server 2 functional modules comprise an engineering management module 20, a neural network model administration module 21, one self-defined node figure generation module 22, a training data administration module 23, a data analysis management module 24, an analysis result administration module 25.Wherein engineering management module 20 comprises the inquiry, management of each engineering information under the engineering group and related with neural network model etc.; Neural network model administration module 21 comprises related datas such as the foundation, deletion, computing of neural network model; Self-defined node figure generation module 22 comprises the drafting of single graphical nodes or the batch duplicating mechanism of same layer pattern node, the automatic matching mechanism of the association line of the node of different layers and layer is (as: when many client computers 4 increase by an input layer in neural network model, the association line of neural network model under then this node and adjacent hiding all node of layer are set up automatically and met), thus realize that the self-defined node figure of neural network model generates and revise all relevant type and parameter informations of model therewith; Training data administration module 23 comprises and related datas such as the foundation of the corresponding sample data that is used to train of neural network model, deletion, computing and the operational administrative that carries out neural metwork training; Data analysis management module 24 comprises and related datas such as the corresponding foundation that is used to analyze data of neural network model, deletion, computing and the operational administrative that carries out analysis of neural network; Analysis result administration module 25 provides the management to the related datas such as object information of analysis of neural network, realizes the output with Suresh Kumar of gathering to analysis result.
As shown in Figure 3, be the self-defined artificial nerve network model hoist pennants of the artificial neural network analysis system and method for a kind of self-definition model of the present invention.The relevant data of the self-defined various neural network model of neural network model table 5 record of database server 1 storage.Self-defined neural network model table 5 comprises following field: sequence number 50, job number 51, model name 52, neural network type 53, the network number of plies 54, kernel function type 55, error precision 56, learning rate 57.Wherein sequence number 50, write down the numbering of the historical record of each neural network model; The numbering of the engineering name record under job number 51 each neural network model; Model name 52, the title of record neural network model; Neural network type 53, the type information under the record neural network model; The network number of plies 54, the hierachy number of record neural network model; Kernel function type 55, the type of the effect kernel function of record neural network model; Error precision 56, the error precision value that the record neural network model will reach; Learning rate 57, record neural network model learning rate parameter.
As shown in Figure 4, be the analysis process figure of the artificial neural network analysis system and method for a kind of self-definition model of the present invention.The artificial neural network analysis system 6 of self-definition model is applied in the working environment of single server to multiple client computer, connects multi-level system platform simultaneously, possesses favorable expansibility and ease for use.The artificial neural network analysis system 6 of self-definition model receives the figure neural network model information (step S40) of many client computers 4, obtains complete neural network model information (step S41) by neural network model administration module 21.The artificial neural network analysis system 6 of self-definition model receives many client computer 4 neural network types, parameter and related engineering related datas, and finishes the generation (step S42) of the full detail of neural network model at database server 1 by neural network model administration module 21.The artificial neural network analysis system 6 of self-definition model receives many client computers 4 and is used to the data related data (step S43) of training and analyzing, judge whether to meet associated neural network model (step S44), then do not return the rapid S43 of previous step if meet, then judge whether the data information (step S45) of training if meet associated neural network model, if training data type, then the artificial neural network analysis system 6 of self-definition model uses and finishes the training computing (step S46) of neural network and return step S43 by training data administration module 23, if the analysis data type, the artificial neural network analysis system 6 of self-definition model uses and finishes data analysis (step S47) by data analysis management module 24.The interpretation of result data that the artificial neural network analysis system 6 of self-definition model uses by analysis result administration module 25 turn back to many client computers 4 (step S48).
The present invention can be of wide application, and the change that is not limited in this instructions to be mentioned does not exceed embodiments of the present invention and field.For the professional in present technique field, any improvement or change that the embodiment of the invention is done do not exceed the protection domain of spirit of the present invention and institute's accessory claim.
Claims (11)
1. artificial neural network analysis system and method based on self-definition model, adopt database server, application server and a plurality of client computer multi-layer framework, provide a kind of method of complete self-defined modeling to realize the quick foundation and the realization of artificial neural network analysis model are carried out comprehensive analyses and prediction to data, it is characterized in that, wherein:
Database server is used to store and various dissimilar relevant model parameter, training sample data and the related data data such as analysis data and analyses and prediction result of neural network model;
Application server is used to receive the neural network model information from client computer, according to neural network model model information and parameter, generates the relevant information of self-defined neural network model, and this application server includes:
One engineering management module is used for the inquiry, management of each engineering information under the engineering group and related with neural network model etc.;
One neural network model administration module is used for according to the self-defining neural network model of client computer, and related datas such as the foundation of neural network model, deletion, computing are managed.
Client computer can be realized the foundation of graphical nodes neural network model, the inquiry of model parameter, setting, modification and deletion or the like and data training and the analysis of carrying out neural network.
2. the artificial neural network analysis system based on self-definition model as claimed in claim 1, it is characterized in that, application server also includes a self-defined node figure generation module, be used for the drafting of single graphical nodes or the batch duplicating mechanism of same layer pattern node, the automatic matching mechanism of the association line of the node of different layers and layer is (as: when many client computers 4 increase by an input layer in neural network model, the association line of neural network model under then this node and adjacent hiding all node of layer are set up automatically and met), thus realize that the self-defined node figure of neural network model generates and revise all relevant type and parameter informations of model therewith.
3. the artificial neural network analysis system based on self-definition model as claimed in claim 1, it is characterized in that, application server also includes a training data administration module, is used for and related datas such as the foundation of the corresponding sample data that is used to train of neural network model, deletion, computing and the operational administrative that carries out neural metwork training.
4. the artificial neural network analysis system based on self-definition model as claimed in claim 1, it is characterized in that, application server also comprises a data analysis management module, is used for and related datas such as the corresponding foundation that is used to analyze data of neural network model, deletion, computing and the operational administrative that carries out analysis of neural network.
5. the artificial neural network analysis system based on self-definition model as claimed in claim 1, it is characterized in that, application server also comprises an analysis result administration module, be used to provide management, realize the output with Suresh Kumar of gathering analysis result to the related datas such as object information of analysis of neural network.
6. the artificial neural network analysis system based on self-definition model as claimed in claim 1, it is characterized in that, database server also comprises the self-defined neural network model table of a storage, be used to write down the relevant data of various neural network model, to guarantee related data records such as all neural network models add at any time, modification, deletion.
7. the artificial neural network analysis method based on self-definition model by the working environment of single server to multiple client computer, connects multi-level system platform and analyzes operation, it is characterized in that, comprising:
(a) the figure neural network model information of many client computers of reception;
(b) obtain complete neural network model information, and the result is deposited in database server;
(c) the data related data that receives training and analyze, judge whether to meet associated neural network model, do not return if meet then, then judge whether the data information of training if meet associated neural network model, if the training data type is then trained computing and is returned, otherwise then analyzes computing;
(d) generate corresponding analysis result information according to neural network model information and analysis data, and the result is deposited in the database server;
(e) realize the output with Suresh Kumar of gathering to analysis result.
8. the artificial neural network analysis method based on self-definition model as claimed in claim 7, it is characterized in that, step (b) comprises reception neural network type, parameter and related engineering related data, and finish the generation of the full detail of neural network model, and be stored in the step of database server by the neural network model administration module.
9. the artificial neural network analysis method based on self-definition model as claimed in claim 7, it is characterized in that, step (c) comprises that neural network model finishes training by training data administration module and data analysis management module or analyze, if train, then all training results are stored into the step of database server.
10. the artificial neural network analysis method based on self-definition model as claimed in claim 7, it is characterized in that, step (d) comprises and related datas such as the corresponding foundation that is used to analyze data of neural network model, deletion, computing and the information management of carrying out analysis of neural network, and is stored in the step of database server.
11. the artificial neural network analysis method based on self-definition model as claimed in claim 7 is characterized in that step (e) comprises the management to the related datas such as object information of analysis of neural network, realizes the output with Suresh Kumar of gathering to analysis result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200610054283 CN101071412A (en) | 2006-05-10 | 2006-05-10 | Neural network analysis system and method based on self-definition model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200610054283 CN101071412A (en) | 2006-05-10 | 2006-05-10 | Neural network analysis system and method based on self-definition model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101071412A true CN101071412A (en) | 2007-11-14 |
Family
ID=38898640
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200610054283 Pending CN101071412A (en) | 2006-05-10 | 2006-05-10 | Neural network analysis system and method based on self-definition model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101071412A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101809539A (en) * | 2007-08-06 | 2010-08-18 | 枢科技术有限责任公司 | Define the method and apparatus of artificial brain via a plurality of concept nodes that link together by predetermined relationship |
| CN102737278A (en) * | 2011-03-31 | 2012-10-17 | 微软公司 | Deep convex network with joint use of nonlinear random projection, restricted boltzmann machine and batch-based parallelizable optimization |
| CN106779583A (en) * | 2016-11-25 | 2017-05-31 | 哈尔滨工程大学 | Extra large frock based on artificial neural network is for project work STRUCTURE DECOMPOSITION System and method for |
| CN108573307A (en) * | 2018-03-05 | 2018-09-25 | 维沃移动通信有限公司 | A kind of method and terminal of processing neural network model file |
| CN108763713A (en) * | 2018-05-22 | 2018-11-06 | 中国电建集团中南勘测设计研究院有限公司 | A kind of mathematical model computational methods and system |
| CN109146057A (en) * | 2018-06-26 | 2019-01-04 | 杭州雄迈集成电路技术有限公司 | A kind of high-precision neural network engineering method based on computation of table lookup |
| CN109726818A (en) * | 2018-12-29 | 2019-05-07 | 北京航天数据股份有限公司 | A kind of model editing method, apparatus, equipment and medium |
| CN109976887A (en) * | 2017-12-28 | 2019-07-05 | 北京中科寒武纪科技有限公司 | Dispatching method and relevant apparatus |
| CN110222877A (en) * | 2019-05-17 | 2019-09-10 | 华中科技大学 | A kind of load prediction system and load forecasting method based on customized neural network |
| CN111144561A (en) * | 2018-11-05 | 2020-05-12 | 杭州海康威视数字技术股份有限公司 | Neural network model determining method and device |
| CN113240114A (en) * | 2021-06-09 | 2021-08-10 | 青岛科技大学 | Artificial neural network-based biomass transformation key variable screening method |
-
2006
- 2006-05-10 CN CN 200610054283 patent/CN101071412A/en active Pending
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101809539A (en) * | 2007-08-06 | 2010-08-18 | 枢科技术有限责任公司 | Define the method and apparatus of artificial brain via a plurality of concept nodes that link together by predetermined relationship |
| CN102737278A (en) * | 2011-03-31 | 2012-10-17 | 微软公司 | Deep convex network with joint use of nonlinear random projection, restricted boltzmann machine and batch-based parallelizable optimization |
| CN102737278B (en) * | 2011-03-31 | 2019-02-12 | 微软技术许可有限责任公司 | The projection of joint non-linear stochastic, limited Boltzmann machine and based on batch can parallel optimization come using deep convex network |
| CN106779583A (en) * | 2016-11-25 | 2017-05-31 | 哈尔滨工程大学 | Extra large frock based on artificial neural network is for project work STRUCTURE DECOMPOSITION System and method for |
| CN109976887A (en) * | 2017-12-28 | 2019-07-05 | 北京中科寒武纪科技有限公司 | Dispatching method and relevant apparatus |
| CN109976887B (en) * | 2017-12-28 | 2020-03-24 | 中科寒武纪科技股份有限公司 | Scheduling method and related device |
| CN108573307A (en) * | 2018-03-05 | 2018-09-25 | 维沃移动通信有限公司 | A kind of method and terminal of processing neural network model file |
| CN108763713A (en) * | 2018-05-22 | 2018-11-06 | 中国电建集团中南勘测设计研究院有限公司 | A kind of mathematical model computational methods and system |
| CN108763713B (en) * | 2018-05-22 | 2022-02-08 | 中国电建集团中南勘测设计研究院有限公司 | Mathematical model calculation method and system |
| CN109146057A (en) * | 2018-06-26 | 2019-01-04 | 杭州雄迈集成电路技术有限公司 | A kind of high-precision neural network engineering method based on computation of table lookup |
| CN109146057B (en) * | 2018-06-26 | 2020-12-08 | 杭州雄迈集成电路技术股份有限公司 | High-precision neural network engineering method based on table lookup calculation |
| CN111144561A (en) * | 2018-11-05 | 2020-05-12 | 杭州海康威视数字技术股份有限公司 | Neural network model determining method and device |
| CN111144561B (en) * | 2018-11-05 | 2023-05-02 | 杭州海康威视数字技术股份有限公司 | Neural network model determining method and device |
| CN109726818A (en) * | 2018-12-29 | 2019-05-07 | 北京航天数据股份有限公司 | A kind of model editing method, apparatus, equipment and medium |
| CN110222877A (en) * | 2019-05-17 | 2019-09-10 | 华中科技大学 | A kind of load prediction system and load forecasting method based on customized neural network |
| CN110222877B (en) * | 2019-05-17 | 2021-08-03 | 华中科技大学 | Load prediction system and load prediction method based on user-defined neural network |
| CN113240114A (en) * | 2021-06-09 | 2021-08-10 | 青岛科技大学 | Artificial neural network-based biomass transformation key variable screening method |
| CN113240114B (en) * | 2021-06-09 | 2023-04-18 | 青岛科技大学 | Artificial neural network-based biomass transformation key variable screening method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101071412A (en) | Neural network analysis system and method based on self-definition model | |
| US8340995B2 (en) | Method and system of using artifacts to identify elements of a component business model | |
| Kwon et al. | Best performance modeling using complementary DEA-ANN approach: Application to Japanese electronics manufacturing firms | |
| Kusiak et al. | A structured approach for analysis of design processes | |
| Aghimien et al. | A review of the application of data mining for sustainable construction in Nigeria | |
| El‐Ghandour et al. | Survey of information technology applications in construction | |
| Relich et al. | Knowledge discovery in enterprise databases for forecasting new product success | |
| Li et al. | Stiff: A forecasting framework for spatiotemporal data | |
| Pandya | Review of modelling techniques and tools for decision making in manufacturing management | |
| WO2005103983A2 (en) | Generative investment process | |
| Uskenbayeva et al. | An algorithm for creating an automated system based on platform of business process | |
| US20230047612A1 (en) | Computer-based system using neuron-like representation graphs to create knowledge models for computing semantics and abstracts in an interactive and automatic mode | |
| Bohács et al. | Production logistics simulation supported by process description languages | |
| Shayakhmetova et al. | Descriptive big data analytics in the field of education | |
| US20140149186A1 (en) | Method and system of using artifacts to identify elements of a component business model | |
| CN104765602B (en) | The Quantitatively Selecting method of nonfunction requirement implementation strategy | |
| Lee et al. | Using workflow for reconfigurable simulation-based planning and scheduling system | |
| KR100743150B1 (en) | Customized and Automated Technology Roadmapping System | |
| Sunkle et al. | Incorporating directives into enterprise TO-BE architecture | |
| Borges et al. | Traffic light control and machine learning: A systematic mapping review | |
| Cisterna et al. | Synergies Between Lean Construction and Artificial Intelligence: AI Driven Continuous Improvement Process | |
| Babu et al. | ANP-GP approach for selection of software architecture styles | |
| Sushil | Physical system theory: fundamentals, recent developments and relationships with system dynamics | |
| Ra̧czka et al. | ALMM solver-database structure and data access layer architecture | |
| Yin | Design and Implementation of Accounting Information System Based on Decision Tree Classification Algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C57 | Notification of unclear or unknown address | ||
| DD01 | Delivery of document by public notice |
Addressee: He Qianjun Document name: Correction notice |
|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |