CN109063846B - Machine learning operation method, device, equipment and storage medium - Google Patents
Machine learning operation method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN109063846B CN109063846B CN201810857947.2A CN201810857947A CN109063846B CN 109063846 B CN109063846 B CN 109063846B CN 201810857947 A CN201810857947 A CN 201810857947A CN 109063846 B CN109063846 B CN 109063846B
- Authority
- CN
- China
- Prior art keywords
- training
- learning
- test set
- configuration
- configuration information
- 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.)
- Active
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Debugging And Monitoring (AREA)
Abstract
The invention discloses an operation method, an apparatus, a device and a storage medium for machine learning, wherein the operation method comprises the following steps: determining that the current configuration information of the learning task is updated; identifying a learning module affected by the update of the current configuration information; running the identified learning module during a current machine learning process of the learning task. The invention effectively solves the problem of resource waste caused by repeatedly operating some learning modules for many times in the machine learning process, and does not need personnel to judge whether the corresponding learning module is operated in the subsequent process in the learning process, thereby effectively reducing the participation degree of the personnel and further effectively reducing the complexity of the machine learning.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a machine learning operation method, a machine learning operation device, machine learning equipment and a storage medium.
Background
Machine Learning (ML) generally includes modules such as sampling of a training set and a test set, feature extraction of the training set and the test set, training of the training set and the test set, and model evaluation of the training set and the test set. The sampling is to extract data according to different data sets required by training and testing and control the number of samples; the feature extraction is to extract the required features for the model of machine learning and perform discretization and other processing on the features; training is to predict the corresponding model from the data according to an expected algorithm; model evaluation is the provision of the behavior of the model on the respective data set.
Therefore, a complete machine learning process is performed in the following order: sampling of a training set, feature extraction of the training set, training of the training set, model evaluation of the training set, sampling of a test set, feature extraction of the test set, training of the test set, and model evaluation of the test set. It can be seen that there are corresponding dependencies among the modules, so if one of the links changes, the subsequent modules need to be run to ensure the integrity of the final model evaluation data.
In an operation mode supported by the prior art, if a final model evaluation result is output, each module needs to be operated independently each time, manual operation needs to be performed for many times, and the problem of failure of an operation sequence cannot be guaranteed; meanwhile, if all the devices are operated each time, the resources are wasted.
Disclosure of Invention
In order to overcome the above-mentioned drawbacks, the present invention provides a method, an apparatus, a device and a storage medium for machine learning, which are used to at least solve the problem of resource waste in the existing machine learning process.
In order to solve the above technical problem, an operation method for machine learning according to an embodiment of the present invention includes:
determining that the current configuration information of the learning task is updated;
identifying a learning module affected by an update of the current configuration information;
running the identified learning module during a current machine learning process of the learning task.
Optionally, the identifying a learning module affected by the update of the current configuration information includes:
and identifying the learning module influenced by the current updating of the configuration information according to the influence relationship between the preset updating of the configuration information and the learning module.
Optionally, the configuration information includes at least one of: data source configuration, training set start-stop date configuration, test set start-stop date configuration, training set sampling configuration, test set sampling configuration, feature configuration and training parameter configuration;
the learning module includes at least one of: sampling a training set, extracting the characteristics of the training set, training the training set, evaluating a model of the training set, sampling a test set, extracting the characteristics of the test set, training the test set and evaluating the model of the test set; the training set is a data set extracted from a data source and used for model training, and the test set is a data set extracted from a data source and used for model testing.
Optionally, the identifying, according to an influence relationship between preset updating of configuration information and a learning module, that the learning module influenced by the updating of the current configuration information is preceded by:
setting influence relations between the data source configuration and all learning modules;
setting influence relations between the start-stop date configuration and the sampling configuration of the training set and the sampling of the training set, the feature extraction of the training set, the training of the training set, the model evaluation of the training set and the model evaluation of the test set;
setting influence relations between the test set start-stop date configuration and the test set sampling configuration and the sampling of the test set, the feature extraction of the test set, the training of the test set and the model evaluation of the test set;
setting influence relations between the feature configurations and feature extraction of the training set, training of the training set, model evaluation of the training set, feature extraction of the test set, training of the test set and model evaluation of the test set;
and setting influence relations between the training configuration and the training of the training set, the model evaluation of the training set, the training of the test set and the model evaluation of the test set.
Optionally, the determining that the current configuration information of the learning task is updated includes:
comparing the update time of the current configuration information with the running time of the learning task in the previous machine learning process;
determining that the current configuration information is updated when the update time changes relative to the runtime.
Optionally, the running the identified learning module in the current machine learning process of the learning task includes:
and in the current machine learning process of the learning task, operating the identified learning modules according to the dependency relationship among the identified learning modules.
In order to solve the above technical problem, an operating device for machine learning according to an embodiment of the present invention includes:
the judging module is used for determining that the current configuration information of the learning task is updated;
an identification module for identifying a learning module affected by the update of the current configuration information;
and the execution module is used for operating the identified learning module in the current machine learning process of the learning task.
Optionally, the identifying module is specifically configured to identify, according to an influence relationship between preset updating of configuration information and a learning module, the learning module influenced by the updating of the current configuration information.
Optionally, the configuration information includes at least one of: data source configuration, training set start-stop date configuration, test set start-stop date configuration, training set sampling configuration, test set sampling configuration, feature configuration and training parameter configuration;
the learning module includes at least one of: sampling a training set, extracting the characteristics of the training set, training the training set, evaluating a model of the training set, sampling a test set, extracting the characteristics of the test set, training the test set and evaluating the model of the test set; the training set is a data set extracted from a data source and used for model training, and the test set is a data set extracted from a data source and used for model testing.
Optionally, the apparatus further comprises: the setting module is used for setting the influence relationship between the data source configuration and all the learning modules; setting influence relations between the start-stop date configuration and the sampling configuration of the training set and the sampling of the training set, the feature extraction of the training set, the training of the training set, the model evaluation of the training set and the model evaluation of the test set; setting influence relations between the test set start-stop date configuration and the test set sampling configuration and the sampling of the test set, the feature extraction of the test set, the training of the test set and the model evaluation of the test set; setting influence relations between the feature configurations and feature extraction of the training set, training of the training set, model evaluation of the training set, feature extraction of the test set, training of the test set and model evaluation of the test set; and setting an influence relationship between the training configuration and the training of the training set, the model evaluation of the training set, the training of the test set and the model evaluation of the test set.
Optionally, the determining module is specifically configured to compare the update time of the current configuration information with the running time of the learning task in the previous machine learning process; determining that the current configuration information is updated when the update time changes relative to the runtime.
Optionally, the executing module is specifically configured to run the identified learning modules according to the dependency relationships between the identified learning modules in the current machine learning process of the learning task.
In order to solve the above technical problem, an apparatus in an embodiment of the present invention includes a memory storing a machine-learning running computer program and a processor executing the computer program to implement the steps of the method according to any one of the above.
To solve the above technical problem, a computer-readable storage medium in an embodiment of the present invention stores a running computer program for machine learning, and the computer program is executable by at least one processor to implement the steps of the method as described in any one of the above.
The invention has the following beneficial effects:
in the embodiments of the invention, the identified operation module is directly operated after the configuration information is changed each time, so that the problem of resource waste caused by repeated operation of some learning modules for many times in the machine learning process is effectively solved, and in the learning process, personnel are not needed to judge whether the corresponding learning module is operated in the subsequent process, thereby effectively reducing the participation degree of the personnel and effectively reducing the complexity of the machine learning.
The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
FIG. 1 is a main flow diagram of a method of operating machine learning in an embodiment of the present invention;
FIG. 2 is a flow chart of an alternative method of operation of machine learning in an embodiment of the present invention;
FIG. 3 is a diagram illustrating influence relationships in an embodiment of the present invention;
FIG. 4 is a flow chart of another method of operation of an alternative machine learning embodiment of the present invention;
FIG. 5 is a schematic diagram of an operating apparatus for machine learning according to an embodiment of the present invention;
fig. 6 is a schematic structural diagram of an apparatus in an embodiment of the invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
In the following description, suffixes such as "module", "component", or "unit" used to denote elements are used only for facilitating the explanation of the present invention, and have no specific meaning in itself. Thus, "module", "component" or "unit" may be used mixedly.
The use of prefixes such as "first," "second," etc. to distinguish between elements is merely intended to facilitate the description of the invention and has no particular meaning in and of themselves.
Example one
An embodiment of the present invention provides an operation method for machine learning, as shown in fig. 1, the method includes:
s101, determining that the current configuration information of the learning task is updated;
s102, identifying a learning module influenced by the update of the current configuration information;
s103, operating the identified learning module in the current machine learning process of the learning task.
In embodiments of the present invention, a learning task may be a process of machine learning for a certain project.
In the embodiment of the invention, the current configuration information refers to the configuration information corresponding to the current machine learning process.
In the embodiment of the invention, the configuration information can comprise data source configuration, training set start-stop date configuration, test set start-stop date configuration, training set sampling configuration, test set sampling configuration, feature configuration and training parameter configuration; wherein, the data source can be stored according to date and is used for the data for machine learning. Training sets and testing sets are part of a data source, wherein the training sets are data sets extracted from the data source and used for model training, the testing sets are data sets extracted from the data source and used for model testing, the dates of the training sets and the testing sets can be predefined, the training sets and the testing sets have no common date, namely, no common data, and the days of the training sets are generally larger than the days of the testing sets; the data source may include various types of business data, etc., with start and end dates representing start and end times. Of course, the configuration information may also include some other custom configurations, such as model file locations on which the fusion model depends, and the like, and if the custom configuration is changed, all learning modules may be run, or corresponding learning modules may be determined according to actual situations.
In an embodiment of the present invention, the learning module may include sampling of a training set, feature extraction of a training set, training of a training set, model evaluation of a training set, sampling of a test set, feature extraction of a test set, training of a test set, and model evaluation of a test set.
According to the embodiment of the invention, when the current configuration information of the learning task is determined to be updated, the learning module influenced by the update of the current configuration information is identified, and the identified learning module is operated in the current machine learning process of the learning task, so that the identified operation module is directly operated after the configuration information is changed every time, and the problem of resource waste caused by repeated operation of some learning modules in the machine learning process is effectively solved, and personnel are not needed to judge whether the corresponding learning module is operated in the subsequent process in the learning process, so that the complexity of machine learning is effectively reduced.
In some embodiments, the learning module that identifies the influence of the update of the current configuration information may also include:
and identifying the learning module influenced by the current updating of the configuration information according to the influence relationship between the preset updating of the configuration information and the learning module.
The machine learning method comprises the steps of identifying through a preset influence relation, and effectively improving the running speed of machine learning.
In some embodiments, the determining that the current configuration information of the learning task is updated may also include:
comparing the update time of the current configuration information with the running time of the learning task in the previous machine learning process;
determining that the current configuration information is updated when the update time changes relative to the runtime.
In some embodiments, the running the identified learning module during the current machine learning process of the learning task may also include:
and in the current machine learning process of the learning task, operating the identified learning modules according to the dependency relationship among the identified learning modules. Wherein the dependency relationship may be a predetermined sequential relationship of back and forth operation between the respective learning modules.
The identified learning modules are operated through the dependency relationship, so that the participation of personnel in the machine learning process can be effectively reduced, and the machine learning complexity is further effectively reduced.
Example two
An embodiment of the present invention provides an operation method for machine learning, as shown in fig. 2, the method includes:
s201, determining that the current configuration information of the learning task is updated;
s202, identifying a learning module influenced by the current configuration information according to the influence relationship between the preset configuration information and the learning module;
s203, in the current machine learning process of the learning task, operating the identified learning module.
According to the embodiment of the invention, the affected learning modules can be quickly identified according to the preset influence relationship, so that the machine learning efficiency can be effectively improved, the learning modules need not to be selected to operate in each operation process of the learning task, the operation sequence of each learning module does not need to be considered, and the learning modules affected by the change of the configuration information can only be operated, so that the problem of resource waste caused by the fact that all the learning modules need to be operated once every time a small configuration modification is made in the prior art can be effectively solved.
In some embodiments, the configuration information comprises at least one of: data source configuration, training set start-stop date configuration, test set start-stop date configuration, training set sampling configuration, test set sampling configuration, feature configuration and training parameter configuration;
the learning module includes at least one of: sampling a training set, extracting the characteristics of the training set, training the training set, evaluating a model of the training set, sampling a test set, extracting the characteristics of the test set, training the test set and evaluating the model of the test set; the training set is a data set extracted from a data source and used for model training, and the test set is a data set extracted from a data source and used for model testing.
In some embodiments, the identifying, according to an influence relationship between the preset update of the configuration information and the learning module, that the learning module influenced by the update of the current configuration information is preceded by:
setting influence relations between the data source configuration and all learning modules;
setting influence relations between the start-stop date configuration and the sampling configuration of the training set and the sampling of the training set, the feature extraction of the training set, the training of the training set, the model evaluation of the training set and the model evaluation of the test set;
setting influence relations between the test set start-stop date configuration and the test set sampling configuration and the sampling of the test set, the feature extraction of the test set, the training of the test set and the model evaluation of the test set;
setting influence relations between the feature configurations and feature extraction of the training set, training of the training set, model evaluation of the training set, feature extraction of the test set, training of the test set and model evaluation of the test set;
setting an influence relationship between the training configuration and the training of the training set, the model evaluation of the training set, the training of the test set, and the model evaluation of the test set.
For example, as shown in fig. 3, when the configuration of the data source is changed, all data need to be recalculated, and all learning modules are operated at this time; when the start-stop date and the sampling configuration of the training set are changed, only the source data of the training set is changed, all modules of the training set can be operated at the moment, and then the model evaluation module of the test set is operated; when the start-stop date of the test set and the sampling configuration of the test set are changed, all modules of the test set only need to be operated, and the model evaluation result of the test set can be obtained; when the feature configuration is changed, the training set and the test set need to be operated with feature extraction, training and model evaluation; when the training parameters are changed, the training set and the test set both need to be run for training and model evaluation.
The embodiment of the invention can effectively solve the problem of resource waste in the machine learning process by the arrangement.
EXAMPLE III
An embodiment of the present invention provides an operation method for machine learning, as shown in fig. 4, the method includes:
s401, comparing the updating time of the current configuration information with the running time of the learning task in the previous machine learning process;
s402, when the updating time is changed relative to the running time, determining that the current configuration information is updated;
s403, determining that the current configuration information of the learning task is updated;
s404, identifying a learning module influenced by the update of the current configuration information;
s405, in the current machine learning process of the learning task, the identified learning module is operated.
That is, in the embodiment of the present invention, the update time of each configuration information and the running time of the previous learning task may be recorded. Then, before each machine learning process of the learning task, the updating time of each configuration information is compared with the last operation time, so that the learning module and the downstream learning module which need to be operated are identified. And finally, providing the parameter type required by operation according to the comparison result of the updating time and the last operation time, operating the ML task, and recording the current operation time so as to identify the module required to be operated in the next operation time.
According to the method provided by the embodiment of the invention, after the configuration information is changed each time, the identified learning module is directly operated, so that the participation of personnel is reduced, and the problem of resource waste of the existing machine learning can be effectively solved.
Example four
An embodiment of the present invention provides an operating device for machine learning, as shown in fig. 5, the device includes:
the judging module 12 is configured to determine that current configuration information of the learning task is updated;
an identification module 14 for identifying a learning module affected by the update of the current configuration information;
an execution module 16, configured to run the identified learning module during a current machine learning process of the learning task.
In embodiments of the present invention, a learning task may be a process of machine learning for a certain project.
In the embodiment of the invention, the current configuration information refers to the configuration information corresponding to the current machine learning process.
In the embodiment of the invention, the configuration information can comprise data source configuration, training set start-stop date configuration, test set start-stop date configuration, training set sampling configuration, test set sampling configuration, feature configuration and training parameter configuration; wherein, the data source can be stored according to date and used for the data for machine learning. The training set and the test set are part of a data source, the dates of the training set and the test set can be defined in advance, the training set and the test set do not have a common date, namely, the training set and the test set do not have common data, and the days of the training set are generally larger than the days of the test set; the data sources may include rented houses, second-hand houses, etc., with start and end dates representing start and end times. Of course, the configuration information may also include some other custom configurations, such as model file locations on which the fusion model depends, and the like, and if the custom configuration is changed, all learning modules may be run, or corresponding learning modules may be determined according to actual situations.
In an embodiment of the present invention, the learning module may include sampling of a training set, feature extraction of a training set, training of a training set, model evaluation of a training set, sampling of a test set, feature extraction of a test set, training of a test set, and model evaluation of a test set.
According to the embodiment of the invention, when the current configuration information of the learning task is determined to be updated, the learning module influenced by the update of the current configuration information is identified, and the identified learning module is operated in the current machine learning process of the learning task, so that the identified operation module is directly operated after the configuration information is changed every time, and the problem of resource waste caused by repeated operation of some learning modules in the machine learning process is effectively solved, and personnel are not needed to judge whether the corresponding learning module is operated in the subsequent process in the learning process, so that the complexity of machine learning is effectively reduced.
In some embodiments, the identifying module 14 is specifically configured to identify a learning module affected by the update of the current configuration information according to an influence relationship between the preset update of the configuration information and the learning module.
The machine learning method has the advantages that the machine learning operation speed can be effectively increased by recognizing the preset influence relationship.
In some embodiments, the determining module 12 is specifically configured to compare the update time of the current configuration information with the running time of the learning task in the previous machine learning process; determining that the current configuration information is updated when the update time changes relative to the runtime.
In some embodiments, the executing module 16 is specifically configured to execute the identified learning modules according to the dependency relationships between the identified learning modules in the current machine learning process of the learning task.
The identified learning modules are operated through the dependency relationship, so that the participation of personnel in the machine learning process can be effectively reduced, and the machine learning complexity is further effectively reduced.
EXAMPLE five
An embodiment of the present invention provides an operating device for machine learning, as shown in fig. 5, the device includes:
the judging module 12 is configured to determine that current configuration information of the learning task is updated;
the identification module 14 is used for identifying the learning module influenced by the update of the current configuration information according to the influence relationship between the preset update of the configuration information and the learning module;
an execution module 16, configured to run the identified learning module during a current machine learning process of the learning task.
According to the embodiment of the invention, the affected learning modules can be quickly identified according to the preset influence relationship, so that the machine learning efficiency can be effectively improved, the learning modules need not to be selected to operate in each operation process of the learning task, the operation sequence of each learning module does not need to be considered, and the learning modules affected by the change of the configuration information can only be operated, so that the problem of resource waste caused by the fact that all the learning modules need to be operated once every time a small configuration modification is made in the prior art can be effectively solved.
In some embodiments, the configuration information comprises at least one of: data source configuration, training set start-stop date configuration, test set start-stop date configuration, training set sampling configuration, test set sampling configuration, feature configuration and training parameter configuration;
the learning module includes at least one of: sampling a training set, extracting the characteristics of the training set, training the training set, evaluating a model of the training set, sampling a test set, extracting the characteristics of the test set, training the test set and evaluating the model of the test set; the training set is a data set extracted from a data source and used for model training, and the test set is a data set extracted from a data source and used for model testing.
In some embodiments, the apparatus may further comprise: the setting module is used for setting the influence relationship between the data source configuration and all the learning modules; setting influence relations between the training set start-stop date configuration and the training set sampling, the training set feature extraction, the training set training, the training set model evaluation and the test set model evaluation; setting influence relations between the test set start-stop date configuration and the test set sampling configuration and the sampling of the test set, the feature extraction of the test set, the training of the test set and the model evaluation of the test set; setting influence relations between the feature configurations and feature extraction of the training set, training of the training set, model evaluation of the training set, feature extraction of the test set, training of the test set and model evaluation of the test set; and setting an influence relationship between the training configuration and the training of the training set, the model evaluation of the training set, the training of the test set and the model evaluation of the test set.
For example, as shown in fig. 3, when the configuration of the data source is changed, all data need to be recalculated, and all learning modules are operated at this time; when the start-stop date and the sampling configuration of the training set are changed, only the source data of the training set is changed, all modules of the training set can be operated at the moment, and then the model evaluation module of the test set is operated; when the start-stop date of the test set and the sampling configuration of the test set are changed, all modules of the test set only need to be operated, and the model evaluation result of the test set can be obtained; when the feature configuration is changed, the training set and the test set need to be operated with feature extraction, training and model evaluation; when the training parameters are changed, the training set and the test set both need to be run for training and model evaluation.
The embodiment of the invention can effectively solve the problem of resource waste in the machine learning process by the arrangement.
Example six
An embodiment of the present invention provides an operating device for machine learning, as shown in fig. 5, the device includes:
a judging module 12, configured to compare the update time of the current configuration information with the previous machine learning process running time of the learning task; determining that the current configuration information is updated when the update time changes relative to the runtime;
an identification module 14 for identifying a learning module affected by the update of the current configuration information;
an execution module 16, configured to run the identified learning module during a current machine learning process of the learning task.
That is, in the embodiment of the present invention, the update time of each configuration information and the running time of the previous learning task may be recorded. Then, before each machine learning process of the learning task, the updating time of each configuration information is compared with the last operation time, so that the learning module and the downstream learning module which need to be operated are identified. And finally, providing the parameter type required by operation according to the comparison result of the updating time and the last operation time, operating the ML task, and recording the current operation time so as to identify the module required to be operated in the next operation time.
The apparatuses in the fourth to sixth embodiments correspond to the corresponding method embodiments, and specific details can be found, so that corresponding technical effects are achieved.
EXAMPLE seven
An apparatus according to an embodiment of the present invention is provided, as shown in fig. 5, and the apparatus includes a memory and a processor, where the memory stores a running computer program for machine learning, and the processor executes the computer program to implement the steps of the method according to any one of the first to third embodiments.
The device in the embodiment of the invention can be a terminal device or a server device.
Example eight
The embodiment of the invention provides a computer-readable storage medium, wherein a machine-learning running computer program is stored in the storage medium, and the computer program can be executed by at least one processor to realize the steps of the method according to any one of the first embodiment to the third embodiment.
When the seventh embodiment and the eighth embodiment are specifically implemented, reference may be made to the above embodiments, and corresponding technical effects are achieved.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (14)
1. A method of operating machine learning, the method comprising:
determining that current configuration information of a learning task is updated, wherein the configuration information comprises at least one of: data source configuration, training set start-stop date configuration, test set start-stop date configuration, training set sampling configuration, test set sampling configuration, feature configuration and training parameter configuration;
identifying a learning module affected by the update of the current configuration information;
running the identified learning module during a current machine learning process of the learning task.
2. The method of claim 1, wherein the identifying a learning module affected by the update of the current configuration information comprises:
and identifying the learning module influenced by the current updating of the configuration information according to the influence relationship between the preset updating of the configuration information and the learning module.
3. The method of claim 2,
the learning module includes at least one of: sampling a training set, extracting the characteristics of the training set, training the training set, evaluating a model of the training set, sampling a test set, extracting the characteristics of the test set, training the test set and evaluating the model of the test set; the training set is a data set extracted from a data source and used for model training, and the test set is a data set extracted from a data source and used for model testing.
4. The method of claim 3, wherein the identifying the learning module before which the current configuration information update affects according to the influence relationship between the preset configuration information update and the learning module comprises:
setting influence relations between the data source configuration and all learning modules;
setting influence relations between the start-stop date configuration and the sampling configuration of the training set and the sampling of the training set, the feature extraction of the training set, the training of the training set, the model evaluation of the training set and the model evaluation of the test set;
setting influence relations between the test set start-stop date configuration and the test set sampling configuration and the sampling of the test set, the feature extraction of the test set, the training of the test set and the model evaluation of the test set;
setting influence relations between the feature configurations and feature extraction of the training set, training of the training set, model evaluation of the training set, feature extraction of the test set, training of the test set and model evaluation of the test set;
setting an influence relationship between the training configuration and the training of the training set, the model evaluation of the training set, the training of the test set, and the model evaluation of the test set.
5. The method of any of claims 1-4, wherein determining that current configuration information for a learning task is updated comprises:
comparing the update time of the current configuration information with the running time of the learning task in the previous machine learning process;
determining that the current configuration information is updated when the update time changes relative to the runtime.
6. The method of any one of claims 1-4, wherein running the identified learning module during the current machine learning process of the learning task comprises:
and in the current machine learning process of the learning task, operating the identified learning modules according to the dependency relationship among the identified learning modules.
7. An operating apparatus for machine learning, the apparatus comprising:
the judging module is used for determining that the current configuration information of the learning task is updated, wherein the configuration information comprises at least one of the following: data source configuration, training set start-stop date configuration, test set start-stop date configuration, training set sampling configuration, test set sampling configuration, feature configuration and training parameter configuration;
an identification module for identifying a learning module affected by the update of the current configuration information;
and the execution module is used for operating the identified learning module in the current machine learning process of the learning task.
8. The apparatus according to claim 7, wherein the identifying module is specifically configured to identify the learning module affected by the update of the current configuration information according to an influence relationship between the update of the preset configuration information and the learning module.
9. The apparatus of claim 8,
the learning module includes at least one of: sampling a training set, extracting the characteristics of the training set, training the training set, evaluating a model of the training set, sampling a test set, extracting the characteristics of the test set, training the test set and evaluating the model of the test set; the training set is a data set extracted from a data source and used for model training, and the test set is a data set extracted from a data source and used for model testing.
10. The apparatus of claim 9, wherein the apparatus further comprises:
the setting module is used for setting the influence relationship between the data source configuration and all the learning modules; setting influence relations between the start-stop date configuration and the sampling configuration of the training set and the sampling of the training set, the feature extraction of the training set, the training of the training set, the model evaluation of the training set and the model evaluation of the test set; setting influence relations between the test set start-stop date configuration and the test set sampling configuration and the sampling of the test set, the feature extraction of the test set, the training of the test set and the model evaluation of the test set; setting influence relations between the feature configurations and feature extraction of the training set, training of the training set, model evaluation of the training set, feature extraction of the test set, training of the test set and model evaluation of the test set; and setting an influence relationship between the training configuration and the training of the training set, the model evaluation of the training set, the training of the test set and the model evaluation of the test set.
11. The apparatus according to any one of claims 7 to 10, wherein the determining module is specifically configured to compare the update time of the current configuration information with a running time of the learning task in a previous machine learning process; determining that the current configuration information is updated when the update time changes relative to the runtime.
12. The apparatus according to any one of claims 7 to 10, wherein the executing module is specifically configured to execute the identified learning modules according to dependencies between the identified learning modules during a current machine learning process of the learning task.
13. An apparatus, characterized in that the apparatus comprises a memory storing a machine-learned running computer program and a processor executing the computer program to implement the steps of the method according to any one of claims 1-6.
14. A computer-readable storage medium, in which a machine-learned running computer program is stored, which computer program is executable by at least one processor for implementing the steps of the method according to any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810857947.2A CN109063846B (en) | 2018-07-31 | 2018-07-31 | Machine learning operation method, device, equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810857947.2A CN109063846B (en) | 2018-07-31 | 2018-07-31 | Machine learning operation method, device, equipment and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109063846A CN109063846A (en) | 2018-12-21 |
CN109063846B true CN109063846B (en) | 2022-05-10 |
Family
ID=64831955
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810857947.2A Active CN109063846B (en) | 2018-07-31 | 2018-07-31 | Machine learning operation method, device, equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109063846B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109815991B (en) * | 2018-12-29 | 2021-02-19 | 北京城市网邻信息技术有限公司 | Training method and device of machine learning model, electronic equipment and storage medium |
CN110414689A (en) * | 2019-08-06 | 2019-11-05 | 中国工商银行股份有限公司 | Update method and device on a kind of machine learning model line |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106250986A (en) * | 2015-06-04 | 2016-12-21 | 波音公司 | Advanced analysis base frame for machine learning |
CN106897224A (en) * | 2017-02-27 | 2017-06-27 | 济南浪潮高新科技投资发展有限公司 | A kind of software test method of determining range and device |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140358828A1 (en) * | 2013-05-29 | 2014-12-04 | Purepredictive, Inc. | Machine learning generated action plan |
US10713594B2 (en) * | 2015-03-20 | 2020-07-14 | Salesforce.Com, Inc. | Systems, methods, and apparatuses for implementing machine learning model training and deployment with a rollback mechanism |
-
2018
- 2018-07-31 CN CN201810857947.2A patent/CN109063846B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106250986A (en) * | 2015-06-04 | 2016-12-21 | 波音公司 | Advanced analysis base frame for machine learning |
CN106897224A (en) * | 2017-02-27 | 2017-06-27 | 济南浪潮高新科技投资发展有限公司 | A kind of software test method of determining range and device |
Also Published As
Publication number | Publication date |
---|---|
CN109063846A (en) | 2018-12-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP2991003B1 (en) | Method and apparatus for classification | |
CN105373478B (en) | Automated testing method and system | |
CN109063846B (en) | Machine learning operation method, device, equipment and storage medium | |
CN106155806A (en) | A kind of multi-task scheduling method and server | |
CN109614325B (en) | Method and device for determining control attribute, electronic equipment and storage medium | |
CN111444718A (en) | Insurance product demand document processing method and device and electronic equipment | |
CN107656860A (en) | Performance test methods, apparatus and system | |
CN116226676B (en) | Machine tool fault prediction model generation method suitable for extreme environment and related equipment | |
CN112114903A (en) | File loading method and device | |
CN114693011A (en) | Policy matching method, device, equipment and medium | |
CN117235527A (en) | End-to-end containerized big data model construction method, device, equipment and medium | |
CN111897725A (en) | Automatic test method, medium, equipment and system for middle platform service | |
JP2018018197A (en) | Source code evaluation program | |
CN112699042B (en) | Unit test case generation method and device | |
CN110716778A (en) | Application compatibility testing method, device and system | |
CN113628077B (en) | Method, terminal and readable storage medium for generating non-repeated questions | |
CN112016607B (en) | Error cause analysis method based on deep learning | |
CN112783775B (en) | Special character input testing method and device | |
CN112395119B (en) | Abnormal data processing method, device, server and storage medium | |
CN114547391A (en) | Message auditing method and device | |
CN111400245B (en) | Art resource migration method and device | |
CN114356781A (en) | Software function testing method and device | |
CN112765041A (en) | Game automatic testing method and device and electronic equipment | |
CN112612702A (en) | Automatic testing method and device based on web | |
EP3547144A1 (en) | Structural tests generation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |