CN110163381A - Intelligence learning method and device - Google Patents
Intelligence learning method and device Download PDFInfo
- Publication number
- CN110163381A CN110163381A CN201910343675.9A CN201910343675A CN110163381A CN 110163381 A CN110163381 A CN 110163381A CN 201910343675 A CN201910343675 A CN 201910343675A CN 110163381 A CN110163381 A CN 110163381A
- Authority
- CN
- China
- Prior art keywords
- algorithm
- level
- integrated information
- information
- base learning
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Machine Translation (AREA)
Abstract
The disclosure provides a kind of intelligence learning method and device, is related to information technology field, is able to solve in artificial intelligence learning process, the inflexible problem of the data adaptive and integration mode of integrated study.The specific technical proposal is: obtaining the first integrated information, the first integrated information is used to indicate the logical relation in level-one algorithm between each algorithm node;At least one level-one algorithm is constructed according to the first integrated information and at least one base learning algorithm, a level-one algorithm includes at least one base learning algorithm;The second integrated information is obtained, the second integrated information is used to indicate the logical relation in second level algorithm between each algorithm node;Second level algorithm is constructed according at least one level-one algorithm, second level algorithm includes at least one level-one algorithm.The disclosure learns for artificial intelligence.
Description
Technical field
This disclosure relates to information technology field more particularly to intelligence learning method and device.
Background technique
In artificial intelligence learning process, learning tasks are completed by building and in conjunction with multiple learning algorithms, wherein single
A learning algorithm is also known as base learning algorithm.Currently, according to the generating mode of individual study, integrated learning approach is roughly divided into two
Major class, the first kind, there are strong dependences between base learning algorithm assigns different base learning algorithms according to learning outcome
Different weights, is then combined;Second class strong dependence is not present between base learning algorithm part, according to different sons
Training gets different base learning algorithms.But the mode of integrated study does not adapt to different learning tasks, user also without
Method is flexibly constructed.
Summary of the invention
The embodiment of the present disclosure provides a kind of intelligence learning method and device, is able to solve in artificial intelligence learning process, collection
At the inflexible problem of the data adaptive and integration mode of study, the technical solution is as follows:
According to the first aspect of the embodiments of the present disclosure, a kind of intelligence learning method is provided, this method comprises:
The first integrated information is obtained, the first integrated information is used to indicate the logic in level-one algorithm between each algorithm node
Relationship;
At least one level-one algorithm, a level-one algorithm are constructed according to the first integrated information and at least one base learning algorithm
Including at least one base learning algorithm;
The second integrated information is obtained, the second integrated information is used to indicate the logic in second level algorithm between each algorithm node
Relationship;
Second level algorithm is constructed according at least one level-one algorithm, second level algorithm includes at least one level-one algorithm.
Two layers of algorithm of level-one algorithm and second level algorithm is constructed according to the first integrated information and the second integrated information, so that two
Grade algorithm can be customized according to user demand, and artificial intelligence study is more flexible, and can be according to data unique characteristics
Corresponding model is selected, study precision is higher.
In one embodiment, the first integrated information includes weighted information, weighted information be used to indicate at least one first
The integration mode of class base learning algorithm;At least one level-one is constructed according to the first integrated information and at least one base learning algorithm to calculate
Method, comprising:
It is combined at least one first kind base learning algorithm to form the first level-one algorithm according to weighted information.
In one embodiment, the first integrated information includes characteristic information, characteristic information be used to indicate at least one second
The characteristic value of class base learning algorithm;At least one level-one is constructed according to the first integrated information and at least one base learning algorithm to calculate
Method, comprising:
It is combined at least one second class base learning algorithm to form the second level-one algorithm according to characteristic information, by sample
When notebook data substitutes into the second level-one algorithm, the corresponding second class base learning algorithm of self-adapted call characteristic value carries out sample data
Processing.
In one embodiment, after constructing second level algorithm according at least one level-one algorithm, method further include:
Sample data is obtained, sample data includes at least one sample;
Second level algorithm is trained according to sample data.
In one embodiment, second level algorithm is trained according to sample data, comprising:
Sample data is substituted into each first kind base learning algorithm to be trained;
The integration mode of each first kind base learning algorithm is determined according to the result of each first kind base learning algorithm,
And the first level-one algorithm is updated, the first level-one algorithm is combined by least one first kind base learning algorithm and is formed.
In one embodiment, second level algorithm is trained according to sample data, comprising:
At least one sample is substituted into the corresponding second class base learning algorithm of characteristic value respectively to be trained;
The second level-one algorithm is updated according to the result of each the second class base learning algorithm.
In one embodiment, the first integrated information is obtained, comprising:
The input operation of user is detected, and generates the first integrated information according to testing result.
According to the second aspect of an embodiment of the present disclosure, a kind of intelligence learning device is provided, comprising: first obtains module, the
One building module, second obtain module, the second building module;
Wherein, first module is obtained, for obtaining the first integrated information, the first integrated information is used to indicate in level-one algorithm
Logical relation between each algorithm node;
First building module, for constructing at least one level-one according to the first integrated information and at least one base learning algorithm
Algorithm, a level-one algorithm include at least one base learning algorithm;
Second obtains module, and for obtaining the second integrated information, the second integrated information is used to indicate each in second level algorithm
Logical relation between algorithm node;
Second building module, for constructing second level algorithm according at least one level-one algorithm, second level algorithm includes at least one
A level-one algorithm.
In one embodiment, the first integrated information includes weighted information, weighted information be used to indicate at least one first
The integration mode of class base learning algorithm;
First building module, is also used to be combined at least one first kind base learning algorithm to be formed according to weighted information
First level-one algorithm.
In one embodiment, the first integrated information includes characteristic information, characteristic information be used to indicate at least one second
The characteristic value of class base learning algorithm;
First building module, is also used to be combined at least one second class base learning algorithm to be formed according to characteristic information
Second level-one algorithm, when sample data is substituted into the second level-one algorithm, the corresponding second class base of self-adapted call characteristic value
Algorithm is practised to handle sample data.
In one embodiment, intelligence learning device further include: third obtains module and training module;
Third obtains module, and for obtaining sample data, sample data includes at least one sample;
Training module, for being trained according to sample data to second level algorithm.
In one embodiment, the first acquisition module includes detection unit and information unit;
Detection unit, the input for detecting user operate;
Information unit, for generating the first integrated information according to the testing result of detection unit.
It should be understood that above general description and following detailed description be only it is exemplary and explanatory, not
The disclosure can be limited.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the disclosure
Example, and together with specification for explaining the principles of this disclosure.
Fig. 1 is a kind of flow chart for intelligence learning method that the embodiment of the present disclosure provides;
Fig. 2 is a kind of algorithm building effect diagram that the embodiment of the present disclosure provides;
Fig. 3 is a kind of algorithm building effect diagram that the embodiment of the present disclosure provides;
Fig. 4 is a kind of algorithm building effect diagram that the embodiment of the present disclosure provides;
Fig. 5 is a kind of intelligence learning Method Modeling schematic diagram that the embodiment of the present disclosure provides;
Fig. 6 is a kind of structure chart for intelligence learning device that the embodiment of the present disclosure provides;
Fig. 7 is a kind of structure chart for intelligence learning device that the embodiment of the present disclosure provides;
Fig. 8 is a kind of structure chart for intelligence learning device that the embodiment of the present disclosure provides.
Specific embodiment
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all implementations consistent with this disclosure.On the contrary, they be only with it is such as appended
The example of the consistent device and method of some aspects be described in detail in claims, the disclosure.
The embodiment of the present disclosure provides a kind of intelligence learning method, is applied to intelligence learning device, as shown in Figure 1, Fig. 1 is this
A kind of flow chart for intelligence learning method that open embodiment provides, the intelligence learning method that the embodiment of the present disclosure provides include with
Lower step:
101, the first integrated information is obtained.
First integrated information is used to indicate the logical relation in level-one algorithm between each algorithm node.
In one embodiment, the first integrated information is obtained, comprising: detect the input operation of user, and tie according to detection
Fruit generates the first integrated information.
First integrated information can be user and be input to intelligence learning device by keyboard or touch screen, be also possible to
User is sent to intelligence learning device after inputting in other equipment, the disclosure to this with no restriction.
102, at least one level-one algorithm is constructed according to the first integrated information and at least one base learning algorithm.
One level-one algorithm includes at least one base learning algorithm.It should be noted that a base learning algorithm can be straight
It connects as level-one algorithm, is also possible to multiple base learning algorithms and combines to form a level-one algorithm.
Herein, the building that two specific examples illustrate level-one algorithm is enumerated, certainly, only exemplary illustration herein, not
It represents the disclosure and is confined to this.
In first example, the first integrated information includes weighted information, weighted information be used to indicate at least one first
The integration mode of class base learning algorithm;At least one level-one is constructed according to the first integrated information and at least one base learning algorithm to calculate
Method, comprising:
It is combined at least one first kind base learning algorithm to form the first level-one algorithm according to weighted information.
Based on first example, in a kind of application scenarios, shown in Fig. 2, base learning algorithm may include traditional base
Practise algorithm and integrated base learning algorithm.By taking Boosting as an example, groundwork mechanism are as follows: obtain one based on training set
Base learning algorithm is adjusted the distribution of training sample further according to the effect of base learning algorithm, so that the study of a upper base is calculated
The training sample of method decision error is obtained bigger weight in this training, then, is trained based on sample distribution adjusted
Next base learning algorithm repeats this process, until the base for reaching specified base learning algorithm number, and training being obtained learns
Algorithm is combined, and obtains final learning algorithm, as level-one algorithm.
In second example, the first integrated information includes characteristic information, characteristic information be used to indicate at least one second
The characteristic value of class base learning algorithm;At least one level-one is constructed according to the first integrated information and at least one base learning algorithm to calculate
Method, comprising:
It is combined at least one second class base learning algorithm to form the second level-one algorithm according to characteristic information, by sample
When notebook data substitutes into the second level-one algorithm, the corresponding second class base learning algorithm of self-adapted call characteristic value carries out sample data
Processing.
Based on second example, in a kind of application scenarios, as shown in figure 3, considering model i.e. when generating combined strategy
Applicability, the model for meeting data characteristics is automatically selected, to improve the performance of model.Its basic thought is: according to data
Different mode selects different models using " door " function, to handle different mode, so that as a result tending to optimal.
Its groundwork mechanism are as follows: setting integrated study first specifies " door " type function as base learning algorithm, then according to figure
Composite structure shown in 2 obtains final learning algorithm based on training set.
Wherein, yi(x) it can be the Ensemble Learning Algorithms such as Boosting, Bagging, or plan is combined using Stacking
Ensemble Learning Algorithms slightly, fiIt (x) is after gate function combines bottom Ensemble Learning Algorithms as a result, then to fi(x) again
Selection combination is carried out using gate function, obtains final result.It is found that the process have passed through and combine twice, so referred to as layer
Secondary combination;For gate function, the realization of the basic functions such as logistic regression, neural network can be used.
103, the second integrated information is obtained.
Second integrated information is used to indicate the logical relation in second level algorithm between each algorithm node.
In one embodiment, the second integrated information is obtained, comprising: detect the input operation of user, and tie according to detection
Fruit generates the second integrated information.
Second integrated information can be user and be input to intelligence learning device by keyboard or touch screen, be also possible to
User is sent to intelligence learning device after inputting in other equipment, the disclosure to this with no restriction.
It should be noted that step 103 and step 101-102 do not have an ordinal relation, step 103 can step 101 it
It is preceding or execute later, also it may be performed simultaneously.
104, second level algorithm is constructed according at least one level-one algorithm.
Second level algorithm includes at least one level-one algorithm.
In a kind of application scenarios, as shown in figure 4, base learning algorithm is learned using logistic regression, SVM etc. are basic
The mode of the integrated studies model such as model and Bagging, Boosting mixing is practised, while being realized using level composite structure to base
The combination of learning algorithm obtains final learning algorithm (i.e. second level algorithm).On the one hand expand the possible model space, while certainly
Those meet the model of data characteristics for dynamic selection, thus lift scheme performance.
In one embodiment, after constructing second level algorithm according at least one level-one algorithm, this method further include:
Sample data is obtained, sample data includes at least one sample;Second level algorithm is trained according to sample data.
Corresponding lower first example of step 102 in one embodiment instructs second level algorithm according to sample data
Practice, comprising:
Sample data is substituted into each first kind base learning algorithm to be trained;
The integration mode of each first kind base learning algorithm is determined according to the result of each first kind base learning algorithm,
And the first level-one algorithm is updated, the first level-one algorithm is combined by least one first kind base learning algorithm and is formed.
Corresponding lower second example of step 102 in one embodiment instructs second level algorithm according to sample data
Practice, comprising:
At least one sample is substituted into the corresponding second class base learning algorithm of characteristic value respectively to be trained;According to each
The result of second class base learning algorithm is updated the second level-one algorithm.
It is instructed it should be noted that bringing at least one sample into characteristic value corresponding second class base learning algorithm respectively
Practice, does not need manually to call, at least one sample can adaptively be substituted into according to characteristic value.
The intelligence learning method in conjunction with described in step 101-104, referring to Figure 5, Fig. 5 are that the embodiment of the present disclosure provides
A kind of intelligence learning Method Modeling schematic diagram.As shown in figure 5, including each base learning algorithm in algorithm node, by base
Two one level learning algorithms can be formed by practising algorithm, and two one level learning algorithms are combined by level and constitute two level learning algorithm.
In order to realize efficient distributed modeling process, in such a way that design and operation mutually separate, that is, modeled
When analysis process designs, each node is not immediately performed, but generates a kind of structure of directed acyclic graph, and enforcement engine can be according to figure
The dependence of each node, optimizes graph structure in structure, and then uses parallel distributed to the node of no dependence
It realizes, finally result is merged, complete modeling procedure.Graph structure as shown in Figure 5, two level-one algorithms can be held parallel
It goes, i.e. Voting algorithm in two level-one algorithms can execute in distributed computing framework simultaneously, SVM algorithm
It can also be executed with parallel distributed.
The intelligence learning method that the embodiment of the present disclosure provides realizes unified group to integrated study and other learning algorithms
It closes nested.The hierarchal model mentioned, base learning algorithm can be the Ensemble Learning Algorithms such as Bagging, Boosting;Hybrid guided mode
In type, base learning algorithm can also be the basic studies algorithm such as hybrid integrated learning algorithm and logistic regression, neural network, expand
The big model space.Level combination, is based on level gate function, corresponding model can be automatically selected according to data unique characteristics, solves
The problem of model adaptation of having determined selects realizes the model adaptation according to data characteristics.Hierarchal model expands Model Space
Between, level combination realizes the adaptive of model, can be according to time unique characteristics, optimal model in preference pattern space is real
The promotion to algorithm performance is showed.
The intelligence learning method that the embodiment of the present disclosure provides, constructs one according to the first integrated information and the second integrated information
Grade two layers of algorithm of algorithm and second level algorithm, is customized second level algorithm according to user demand, and artificial intelligence learns more
Add flexibly, and corresponding model can be selected according to data unique characteristics, study precision is higher.
Based on intelligence learning method described in the corresponding embodiment kind of above-mentioned Fig. 1, the embodiment of the present disclosure provides a kind of intelligence
Energy learning device, for executing parking charging method described in the corresponding embodiment of above-mentioned Fig. 1, as shown in fig. 6, the intelligence
Learning device 60 includes: that the first acquisition module 601, first building module 602, second obtains the building module of module 603, second
604;
Wherein, first module 601 is obtained, for obtaining the first integrated information, the first integrated information is used to indicate level-one calculation
Logical relation in method between each algorithm node;
First building module 602, for constructing at least one according to the first integrated information and at least one base learning algorithm
Level-one algorithm, a level-one algorithm include at least one base learning algorithm;
Second obtains module 603, and for obtaining the second integrated information, the second integrated information is used to indicate in second level algorithm respectively
Logical relation between a algorithm node;
Second building module 604, for constructing second level algorithm according at least one level-one algorithm, second level algorithm includes at least
One level-one algorithm.
In one embodiment, the first integrated information includes weighted information, weighted information be used to indicate at least one first
The integration mode of class base learning algorithm;
First building module 602, is also used to be combined at least one first kind base learning algorithm according to weighted information
Form the first level-one algorithm.
In one embodiment, the first integrated information includes characteristic information, characteristic information be used to indicate at least one second
The characteristic value of class base learning algorithm;
First building module 602, is also used to be combined at least one second class base learning algorithm according to characteristic information
The second level-one algorithm is formed, when sample data is substituted into the second level-one algorithm, corresponding second class of self-adapted call characteristic value
Base learning algorithm handles sample data.
In one embodiment, as shown in fig. 7, intelligence learning device 60 further include: third obtains module 605 and training mould
Block 606;
Third obtains module 605, and for obtaining sample data, sample data includes at least one sample;
Training module 606, for being trained according to sample data to second level algorithm.
In one embodiment, as shown in figure 8, the first acquisition module 601 includes detection unit 6011 and information unit
6012;
Detection unit 6011, the input for detecting user operate;
Information unit 6012, for generating the first integrated information according to the testing result of detection unit.
The intelligence learning device that the embodiment of the present disclosure provides, constructs one according to the first integrated information and the second integrated information
Grade two layers of algorithm of algorithm and second level algorithm, is customized second level algorithm according to user demand, and artificial intelligence learns more
Add flexibly, and corresponding model can be selected according to data unique characteristics, study precision is higher.
Based on intelligence learning method described in the corresponding embodiment of above-mentioned Fig. 1, the embodiment of the present disclosure also provides one kind
Computer readable storage medium, for example, non-transitorycomputer readable storage medium can be read-only memory (English: Read
Only Memory, ROM), it is random access memory (English: Random Access Memory, RAM), CD-ROM, tape, soft
Disk and optical data storage devices etc..It is stored with computer instruction on the storage medium, for executing the corresponding embodiment of above-mentioned Fig. 1
Described in intelligence learning method, details are not described herein again.
Those skilled in the art will readily occur to its of the disclosure after considering specification and practicing disclosure disclosed herein
Its embodiment.This application is intended to cover any variations, uses, or adaptations of the disclosure, these modifications, purposes or
Person's adaptive change follows the general principles of this disclosure and including the undocumented common knowledge in the art of the disclosure
Or conventional techniques.The description and examples are only to be considered as illustrative, and the true scope and spirit of the disclosure are by following
Claim is pointed out.
Claims (10)
1. a kind of intelligence learning method, which is characterized in that the described method includes:
The first integrated information is obtained, first integrated information is used to indicate the logic in level-one algorithm between each algorithm node
Relationship;
At least one level-one algorithm, a level-one algorithm are constructed according to first integrated information and at least one base learning algorithm
Including at least one base learning algorithm;
The second integrated information is obtained, second integrated information is used to indicate the logic in second level algorithm between each algorithm node
Relationship;
The second level algorithm is constructed according at least one described level-one algorithm, the second level algorithm includes at least one described level-one
Algorithm.
2. the method according to claim 1, wherein first integrated information includes weighted information, it is described plus
Power information is used to indicate the integration mode of at least one first kind base learning algorithm;According to first integrated information and at least one
A base learning algorithm constructs at least one level-one algorithm, comprising:
It is combined at least one described first kind base learning algorithm to form the first level-one algorithm according to the weighted information.
3. the method according to claim 1, wherein first integrated information includes characteristic information, the spy
Reference ceases the characteristic value for being used to indicate at least one the second class base learning algorithm;According to first integrated information and at least one
Base learning algorithm constructs at least one level-one algorithm, comprising:
It is combined at least one described second class base learning algorithm to form the second level-one algorithm according to the characteristic information,
When sample data is substituted into the second level-one algorithm, the corresponding second class base learning algorithm of self-adapted call characteristic value is to sample
Data are handled.
4. the method according to claim 1, wherein constructing the second level according at least one described level-one algorithm
After algorithm, the method also includes:
Sample data is obtained, the sample data includes at least one sample;
The second level algorithm is trained according to the sample data.
5. method according to claim 1-4, which is characterized in that obtain the first integrated information, comprising:
The input operation of user is detected, and generates first integrated information according to testing result.
6. a kind of intelligence learning device characterized by comprising first obtains module, the first building module, the second acquisition mould
Block, the second building module;
Wherein, described first module is obtained, for obtaining the first integrated information, first integrated information is used to indicate level-one calculation
Logical relation in method between each algorithm node;
The first building module, for constructing at least one according to first integrated information and at least one base learning algorithm
Level-one algorithm, a level-one algorithm include at least one base learning algorithm;
Described second obtains module, and for obtaining the second integrated information, second integrated information is used to indicate in second level algorithm
Logical relation between each algorithm node;
The second building module, for constructing the second level algorithm according at least one described level-one algorithm, the second level is calculated
Method includes at least one described level-one algorithm.
7. according to the method described in claim 6, it is characterized in that, first integrated information includes weighted information, it is described plus
Power information is used to indicate the integration mode of at least one first kind base learning algorithm;
The first building module is also used to be carried out at least one described first kind base learning algorithm according to the weighted information
Combination forms the first level-one algorithm.
8. according to the method described in claim 6, it is characterized in that, first integrated information includes characteristic information, the spy
Reference ceases the characteristic value for being used to indicate at least one the second class base learning algorithm;
The first building module is also used to be carried out at least one described second class base learning algorithm according to the characteristic information
Combination forms the second level-one algorithm, and when sample data is substituted into the second level-one algorithm, self-adapted call characteristic value is corresponding
The second class base learning algorithm sample data is handled.
9. according to the method described in claim 6, it is characterized in that, the intelligence learning device further include: third obtains module
And training module;
The third obtains module, and for obtaining sample data, the sample data includes at least one sample;
The training module, for being trained according to the sample data to the second level algorithm.
10. according to the described in any item devices of claim 6-9, which is characterized in that described first, which obtains module, includes detection list
Member and information unit;
The detection unit, the input for detecting user operate;
The information unit, for generating first integrated information according to the testing result of the detection unit.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910343675.9A CN110163381A (en) | 2019-04-26 | 2019-04-26 | Intelligence learning method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910343675.9A CN110163381A (en) | 2019-04-26 | 2019-04-26 | Intelligence learning method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110163381A true CN110163381A (en) | 2019-08-23 |
Family
ID=67640112
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910343675.9A Pending CN110163381A (en) | 2019-04-26 | 2019-04-26 | Intelligence learning method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110163381A (en) |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106779087A (en) * | 2016-11-30 | 2017-05-31 | 福建亿榕信息技术有限公司 | A kind of general-purpose machinery learning data analysis platform |
CN107423442A (en) * | 2017-08-07 | 2017-12-01 | 火烈鸟网络(广州)股份有限公司 | Method and system, storage medium and computer equipment are recommended in application based on user's portrait behavioural analysis |
CN107507038A (en) * | 2017-09-01 | 2017-12-22 | 美林数据技术股份有限公司 | A kind of electricity charge sensitive users analysis method based on stacking and bagging algorithms |
CN107563731A (en) * | 2017-09-01 | 2018-01-09 | 上海诺悦智能科技有限公司 | A kind of engineering stream based on data analysis builds system |
CN108023876A (en) * | 2017-11-20 | 2018-05-11 | 西安电子科技大学 | Intrusion detection method and intruding detection system based on sustainability integrated study |
CN108090788A (en) * | 2017-12-22 | 2018-05-29 | 苏州大学 | Ad conversion rates predictor method based on temporal information integrated model |
CN108491970A (en) * | 2018-03-19 | 2018-09-04 | 东北大学 | A kind of Predict Model of Air Pollutant Density based on RBF neural |
CN108830328A (en) * | 2018-06-21 | 2018-11-16 | 中国矿业大学 | Merge the microseismic signals SMOTE recognition methods and monitoring system of spatial knowledge |
CN108985335A (en) * | 2018-06-19 | 2018-12-11 | 中国原子能科学研究院 | The integrated study prediction technique of nuclear reactor cladding materials void swelling |
CN109117864A (en) * | 2018-07-13 | 2019-01-01 | 华南理工大学 | Coronary heart disease risk prediction technique, model and system based on heterogeneous characteristic fusion |
CN109145948A (en) * | 2018-07-18 | 2019-01-04 | 宁波沙塔信息技术有限公司 | A kind of injection molding machine putty method for detecting abnormality based on integrated study |
-
2019
- 2019-04-26 CN CN201910343675.9A patent/CN110163381A/en active Pending
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106779087A (en) * | 2016-11-30 | 2017-05-31 | 福建亿榕信息技术有限公司 | A kind of general-purpose machinery learning data analysis platform |
CN107423442A (en) * | 2017-08-07 | 2017-12-01 | 火烈鸟网络(广州)股份有限公司 | Method and system, storage medium and computer equipment are recommended in application based on user's portrait behavioural analysis |
CN107507038A (en) * | 2017-09-01 | 2017-12-22 | 美林数据技术股份有限公司 | A kind of electricity charge sensitive users analysis method based on stacking and bagging algorithms |
CN107563731A (en) * | 2017-09-01 | 2018-01-09 | 上海诺悦智能科技有限公司 | A kind of engineering stream based on data analysis builds system |
CN108023876A (en) * | 2017-11-20 | 2018-05-11 | 西安电子科技大学 | Intrusion detection method and intruding detection system based on sustainability integrated study |
CN108090788A (en) * | 2017-12-22 | 2018-05-29 | 苏州大学 | Ad conversion rates predictor method based on temporal information integrated model |
CN108491970A (en) * | 2018-03-19 | 2018-09-04 | 东北大学 | A kind of Predict Model of Air Pollutant Density based on RBF neural |
CN108985335A (en) * | 2018-06-19 | 2018-12-11 | 中国原子能科学研究院 | The integrated study prediction technique of nuclear reactor cladding materials void swelling |
CN108830328A (en) * | 2018-06-21 | 2018-11-16 | 中国矿业大学 | Merge the microseismic signals SMOTE recognition methods and monitoring system of spatial knowledge |
CN109117864A (en) * | 2018-07-13 | 2019-01-01 | 华南理工大学 | Coronary heart disease risk prediction technique, model and system based on heterogeneous characteristic fusion |
CN109145948A (en) * | 2018-07-18 | 2019-01-04 | 宁波沙塔信息技术有限公司 | A kind of injection molding machine putty method for detecting abnormality based on integrated study |
Non-Patent Citations (1)
Title |
---|
MAYGOD1IKE: "详解stacking过程", 《CSDN》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Stork et al. | A new taxonomy of global optimization algorithms | |
Li et al. | Data-driven evolutionary algorithm with perturbation-based ensemble surrogates | |
Wu et al. | Ensemble strategies for population-based optimization algorithms–A survey | |
WO2021254114A1 (en) | Method and apparatus for constructing multitask learning model, electronic device and storage medium | |
Chen et al. | Case-based reasoning system and artificial neural networks: A review | |
JP2008533615A (en) | Neural network development and data analysis tools | |
CN107209873A (en) | Hyper parameter for depth convolutional network is selected | |
Mairiza et al. | Utilizing TOPSIS: A multi criteria decision analysis technique for non-functional requirements conflicts | |
CN110222848A (en) | The determination method and device for the integrated model that computer executes | |
Carmona et al. | Evolutionary algorithms for subgroup discovery applied to e-learning data | |
CN110738362A (en) | method for constructing prediction model based on improved multivariate cosmic algorithm | |
CN105260746A (en) | Expandable multilayer integrated multi-mark learning system | |
Gaier et al. | Data-efficient neuroevolution with kernel-based surrogate models | |
CN110222838A (en) | Deep neural network and its training method, device, electronic equipment and storage medium | |
Kamalian et al. | Reducing human fatigue in interactive evolutionary computation through fuzzy systems and machine learning systems | |
Prentzas et al. | Combinations of case-based reasoning with other intelligent methods | |
CN115168720A (en) | Content interaction prediction method and related equipment | |
CN117494775A (en) | Method for training neural network model, electronic equipment, cloud, cluster and medium | |
CN101226521A (en) | Machine learning method for ambiguity data object estimation modeling | |
JP6725194B2 (en) | Methods for generating trained models, methods for classifying data, computers and programs | |
CN110163381A (en) | Intelligence learning method and device | |
Sivaraman et al. | Modeling of an inverted pendulum based on fuzzy clustering techniques | |
Ahmad et al. | A novel adaptive learning path method | |
CN115168722A (en) | Content interaction prediction method and related equipment | |
KR20230135838A (en) | Method and apparatus for selective ensemble prediction based on dynamic model combination |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190823 |
|
RJ01 | Rejection of invention patent application after publication |