CN111428869A - Model generation method and device, computer equipment and storage medium - Google Patents
Model generation method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN111428869A CN111428869A CN202010196133.6A CN202010196133A CN111428869A CN 111428869 A CN111428869 A CN 111428869A CN 202010196133 A CN202010196133 A CN 202010196133A CN 111428869 A CN111428869 A CN 111428869A
- Authority
- CN
- China
- Prior art keywords
- neural network
- network model
- training
- sample
- preset
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/061—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using biological neurons, e.g. biological neurons connected to an integrated circuit
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Theoretical Computer Science (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Neurology (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to a method, an apparatus, a computer device and a storage medium for model generation. The method comprises the following steps: the method comprises the steps that computer equipment obtains sample parameters of a neural network model to be created; acquiring a training data sample of the neural network model according to the sample parameters; the method comprises the steps that computer equipment receives a neural network establishing instruction input by a user, and a neural network model is established, wherein the neural network establishing instruction comprises the types of all network layers contained in the neural network model, the activation state of each network layer, the number of contained neurons and input data dimensions; and the computer equipment trains the neural network model according to the training data sample, the training method and the training sample strategy input by the user to obtain the target neural network model. By adopting the method, the generation of the neural network model can be realized in a man-machine interaction mode.
Description
Technical Field
The present application relates to the field of computer technologies, and in particular, to a method and an apparatus for generating a model, a computer device, and a storage medium.
Background
With the advent of the big data era, more and more data processing methods and models are emerging, wherein data processing is performed by using a neural network model, and the data processing method and the data processing models can be widely applied to multiple fields of neuroscience, artificial intelligence, information processing and the like.
However, the generation of the neural network model requires a user to have professional programming capability and professional programming equipment, and therefore, a method for generating the model by simple setup is needed.
Disclosure of Invention
In view of the above, it is necessary to provide a method, an apparatus, a computer device and a storage medium for model generation in order to solve the above technical problems.
In a first aspect, a method for model generation is provided, the method comprising:
acquiring sample parameters of a neural network model to be created;
acquiring a training data sample of the neural network model according to the sample parameter;
receiving a neural network creating instruction input by a user, creating the neural network model, wherein the neural network creating instruction comprises the types of all network layers included in the neural network model, the activation state of each network layer, the number of included neurons and an input data dimension;
and training the neural network model according to the training data sample, the training method and a training sample strategy input by a user to obtain a target neural network model.
As an optional implementation, the method further comprises:
acquiring a model convergence curve of the target neural network model, and judging the convergence of the target neural network model according to a loss value and a metric value in the model convergence curve;
if the loss value is less than the preset loss threshold and the metric value is less than the preset metric threshold, determining that the target neural network model converges;
and if the loss value is greater than or equal to the preset loss threshold value or the metric value is greater than or equal to the preset metric threshold value, judging that the target neural network model does not converge, and prompting a user to modify the training sample strategy and the compiling parameters in the training method again.
As an optional implementation, the method further comprises:
obtaining a verification data set according to a verification splitting parameter in the training sample strategy, and performing verification scoring processing on the target neural network model according to the verification data set to obtain a verification score;
if the verification score is lower than a preset score threshold value, the step of training the neural network model according to the training data sample, the training method and a training sample strategy input by a user is executed again to obtain a target neural network model;
deriving the target neural network model if the verification score is equal to or above the preset score threshold.
As an optional implementation, the obtaining a training data sample of the neural network model according to the sample parameter includes:
acquiring an imported data sample according to the sample parameter;
and according to a preset data processing method, carrying out extraction-conversion-loading processing on the imported data sample to obtain the training data sample.
In a second aspect, an apparatus for generating a model is provided, the apparatus comprising:
the first acquisition module is used for acquiring sample parameters of a neural network model to be created;
the second acquisition module is used for acquiring a training data sample of the neural network model according to the sample parameter;
the creating module is used for receiving a neural network creating instruction input by a user and creating the neural network model, wherein the neural network creating instruction comprises the types of all network layers contained in the neural network model, the activation state of each network layer, the number of contained neurons and the dimension of input data;
and the training module is used for training the neural network model according to the training data sample, the training method and the training sample strategy input by the user to obtain the target neural network model.
As an optional implementation, the apparatus further comprises:
the third acquisition module is used for acquiring a model convergence curve of the target neural network model and judging the convergence of the target neural network model according to the loss value and the metric value in the model convergence curve;
a first determining module, configured to determine that the target neural network model converges if the loss value is less than the preset loss threshold and the metric value is less than the preset metric threshold;
and the second judging module is used for judging that the target neural network model does not converge if the loss value is greater than or equal to the preset loss threshold or the metric value is greater than or equal to the preset metric threshold, and prompting a user to modify the training sample strategy and the compiling parameters in the training method again.
As an optional implementation, the apparatus further comprises:
the scoring module is used for obtaining a verification data set according to the verification splitting parameters in the training sample strategy and carrying out verification scoring processing on the target neural network model according to the verification data set to obtain a verification score;
a third judging module, configured to, if the verification score is lower than a preset score threshold, re-execute the training sample strategy according to the training data sample, the training method, and the user input, and train the neural network model to obtain a target neural network model;
a fourth determining module for deriving the target neural network model if the verification score is equal to or higher than the preset score threshold.
As an optional implementation manner, the second obtaining module is specifically configured to obtain an import data sample according to the sample parameter;
and according to a preset data processing method, carrying out extraction-conversion-loading processing on the imported data sample to obtain the training data sample.
In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the computer program:
acquiring sample parameters of a neural network model to be created;
acquiring a training data sample of the neural network model according to the sample parameter;
receiving a neural network creating instruction input by a user, creating the neural network model, wherein the neural network creating instruction comprises the types of all network layers included in the neural network model, the activation state of each network layer, the number of included neurons and an input data dimension;
and training the neural network model according to the training data sample, the training method and a training sample strategy input by a user to obtain a target neural network model.
In a fourth aspect, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:
acquiring sample parameters of a neural network model to be created;
acquiring a training data sample of the neural network model according to the sample parameter;
receiving a neural network creating instruction input by a user, creating the neural network model, wherein the neural network creating instruction comprises the types of all network layers included in the neural network model, the activation state of each network layer, the number of included neurons and an input data dimension;
and training the neural network model according to the training data sample, the training method and a training sample strategy input by a user to obtain a target neural network model.
The application provides a method, a device, computer equipment and a storage medium for generating a model, wherein the computer equipment acquires sample parameters of a neural network model to be created; acquiring a training data sample of the neural network model according to the sample parameter; then, the computer equipment receives a neural network establishing instruction input by a user, and establishes a neural network model, wherein the neural network establishing instruction comprises the type of each network layer contained in the neural network model, the activation state of each network layer, the number of contained neurons and the dimension of input data; and finally, the computer equipment trains the neural network model according to the training data sample, the training method and the training sample strategy input by the user to obtain a target neural network model. By adopting the method, the generation of the neural network model can be realized in a man-machine interaction mode.
Drawings
Fig. 1 is a schematic flowchart of a method for generating a model according to an embodiment of the present disclosure;
FIG. 2 is a schematic structural diagram of an apparatus for model generation according to an embodiment of the present disclosure;
fig. 3 is an internal structural diagram of a computer device according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The embodiment of the application provides a model generation method, which can be applied to computer equipment, in particular to model generation software in the computer equipment. In the embodiment provided by the application, the model generation method can generate the wind power plant neural network model, so that a user can predict the wind power plant tower top torque load according to the generated wind power plant neural network model. Firstly, obtaining a sample parameter of a neural network model to be created by computer equipment; and obtaining a training data sample of the neural network model according to the sample parameters. Then, the computer equipment receives a neural network establishing instruction input by a user and establishes a neural network model, wherein the neural network establishing instruction comprises the type of each network layer contained in the neural network model, the activation state of each network layer, the number of contained neurons and the dimension of input data; and then, the computer equipment trains the neural network model according to the training data sample, the training method and the training sample strategy input by the user to obtain the target neural network model.
The embodiment of the application provides a method for generating a model, as shown in fig. 1, the specific processing procedures are as follows:
step 101, obtaining sample parameters of a neural network model to be created.
In implementation, a user sends a request for creating a sample parameter through a human-computer interaction interface of a terminal device (e.g., a computer device), and the computer device obtains the sample parameter of the neural network model to be created according to the request for creating the sample parameter. For example, when creating a neural network model for wind farm tower top torque load prediction, the sample parameters may be, but are not limited to: wind farm active power (mean and standard deviation), wind turbine blade 3 pitch angle (mean and standard deviation), wind turbine impeller speed, wind speed, and tower top torque load.
And 102, acquiring a training data sample of the neural network model according to the sample parameter.
In implementation, the computer device obtains a training data sample of the neural network model to be created according to the sample parameters, wherein the training sample comprises data values corresponding to the sample parameters. Specifically, a user may enable the computer device to obtain a training data sample corresponding to the neural network model to be created by importing an Excel form file into the computer device, or the computer device queries and imports a training data sample corresponding to the neural network model to be created in a database according to a sample parameter, so that the embodiment of the present application is not limited.
As an optional implementation manner, according to the sample parameters, obtaining a training data sample of the neural network model to be created, and the specific processing procedure is as follows:
acquiring an imported data sample according to the sample parameter; and according to a preset data processing method, carrying out extraction-conversion-loading processing on the imported data sample to obtain a training data sample.
In an implementation, a computer device acquires an imported data sample according to a sample parameter, and then performs data preprocessing on the imported data sample, namely, Extract-Transform-load (ET L, Extract-Transform-L oad) processing on the imported data sample to obtain a processed training data sample, wherein ET L processing comprises a plurality of data processing modes, specifically, when the computer device performs ET L processing on the imported data sample (such as a wind farm data sample), null processing, normalization processing, invalid data cleaning, repeated data processing and the like can be included, firstly, the computer device traverses all data values in the imported data sample and compares the data values with a preset null value field, if the traversed data is consistent with the preset null value field, determines that the data is null value, and replaces the null value data with a default data value, and meanwhile, the computer device compares the data values corresponding to each sample parameter in the imported data sample one by one and determines whether the data value in the imported data sample exceeds a preset parameter range, if the data value in the sample exceeds the preset null value field, and deletes the data as invalid data according to the computer device, and deletes the data processing algorithm of the data as the computer device, the invalid data,
) The method comprises the steps of normalizing data in imported data samples, then, cleaning repeated data in data values corresponding to each sample parameter in the imported data samples by computer equipment, wherein the data cleaning process is similar to an invalid data cleaning process, and is not repeated in the application.
Step 103, receiving a neural network creating instruction input by a user, creating a neural network model, wherein the neural network creating instruction comprises the types of all network layers included in the neural network model, the activation state of each network layer, the number of included neurons and the dimension of input data.
In implementation, a user edits and creates a neural network model through a human-computer interaction interface of a computer device, namely inputs a neural network creation instruction, and the computer device creates the neural network model after receiving the neural network creation instruction, wherein the neural network creation instruction comprises types of network layers, types of neurons contained in each network layer, an activation state and an input data dimension.
And 104, training the neural network model according to the training data sample, the training method and the training sample strategy input by the user to obtain the target neural network model.
In implementation, according to a corresponding training data sample (e.g., a wind farm data sample), a user may select a training method (e.g., an SGD gradient descent algorithm, a Nadam adaptive algorithm, a newton method, etc.) of a neural network model through a human-computer interaction interface of a computer device, and then, the user compiles a selected compiling parameter in the training method through the computer device, and at the same time, the user inputs a corresponding training sample strategy to train the created neural network model, thereby obtaining a target neural network model.
Optionally, the user selects a gradient descent algorithm (SGD algorithm) as a training method for the wind farm data (training data sample), and the compiling parameters may include a learning rate (learning rate), a learning rate attenuation (learning rate attenuation), an impulse (momentum), and the like, which is not limited in the embodiment of the present application. By adjusting the values of these compiling parameters, the training speed, the training result, and the like of the training method can be adjusted. The compiling parameters all have initial values, and after the computer equipment trains the neural network model according to the initial values of the compiling parameters, a user can adjust the initial values of the compiling parameters according to the obtained model training result so as to adjust the neural network model.
Optionally, the training sample strategy input by the user may include model training parameters, such as: the number parameter of the full data set (epoch), the number parameter of the training samples (batch _ size), the verification split parameter (validation _ split), and the like, which are not limited in the embodiments of the present application.
Specifically, when the computer device trains the neural network model, it needs to configure a training sample strategy of the neural network model in addition to the training method and the compiling parameters included in the training method, and further, the computer device trains the neural network model according to the model training sample strategy received from the user. For example, in the model training process, the computer device needs to iteratively train all data and needs to repeat multiple times to perform fitting convergence, so the computer device needs multiple full data sets (epochs) to update the iterative weights, and therefore, a user needs to set the number parameters of the full data sets in a training sample strategy; and when the computer device iterates each full data set, each time partial data needs to be input, therefore, the full data set needs to be divided into a plurality of training samples (batch), so that a user needs to set the number of samples (batch _ size) in each training sample, finally, when the computer device verifies the trained neural network model, the computer device needs to verify a splitting parameter (splitting _ split), according to the verification splitting parameter, partial data is selected from the original full data set as a verification set and is input to the neural network model, and a model output result is verified.
As an alternative embodiment, the computer device may also graphically display (e.g., as a flowchart) the network hierarchy of the generated target neural network model. Corresponding codes can also be generated by each network layer of the target neural network model.
As an optional implementation manner, a model convergence curve of the target neural network model is obtained, and the convergence of the target neural network model is judged according to a loss value and a metric value in the model convergence curve; if the loss value is smaller than a preset loss threshold value and the metric value is smaller than a preset metric threshold value, determining that the target neural network model is converged; and if the loss value is greater than or equal to the preset loss threshold value or the metric value is greater than or equal to the preset metric threshold value, judging that the target neural network model does not converge, and prompting the user to modify the compiling parameters in the training sample strategy and the training method again.
In implementation, after obtaining the target neural network model, the computer device may obtain a model convergence curve of the corresponding target neural network, and then, the computer device may determine convergence of the target neural network model according to a loss value and a metric value in the model convergence curve. If the loss value is smaller than the preset loss threshold value and the metric value is smaller than the preset metric threshold value, the computer device judges that the target neural network model is converged, namely the training of the target neural network model is completed and the target neural network model can be used. If the loss value is greater than or equal to the preset loss threshold or the metric value is greater than or equal to the preset metric threshold, the computer device determines that the target neural network model does not converge, i.e., the target neural network model cannot. Therefore, the computer device prompts the user to modify the compiling parameters in the training sample strategy and the training method again through the display interface until the target neural network model converges.
As an optional implementation manner, obtaining a verification data set according to a verification splitting parameter in a training sample strategy, and performing verification scoring processing on a target neural network model according to the verification data set to obtain a verification score; if the verification score is lower than the preset score threshold value, re-executing the training sample strategy according to the training data sample, the training method and the user input, and training the neural network model to obtain a target neural network model; if the verification score is equal to or higher than a preset score threshold, a target neural network model is derived.
In implementation, the computer device obtains a verification data set according to the obtained verification splitting parameters in the training sample strategy, and then, the computer device performs verification scoring processing on the obtained target neural network model according to the verification data set to obtain a verification score. If the scoring result is lower than the preset score threshold, the computer device re-executes the step 104; if the scoring result is equal to or higher than a preset score threshold, the computer device derives a target neural network model.
The embodiment of the application provides a model generation method, wherein computer equipment acquires sample parameters of a neural network model to be created; acquiring a training data sample of the neural network model according to the sample parameters; then, the computer equipment receives a neural network establishing instruction input by a user and establishes a neural network model, wherein the neural network establishing instruction comprises the type of each network layer contained in the neural network model, the activation state of each network layer, the number of contained neurons and the dimension of input data; and finally, the computer equipment trains the neural network model according to the training data sample, the training method and the training sample strategy input by the user to obtain the target neural network model. By adopting the method, the generation of the neural network model can be realized in a man-machine interaction mode.
In one embodiment, when the method of model generation is applied to the application scenario of prediction of wind farm tower top torque load, then the sample parameters may include: the active power (mean value and standard deviation) of the wind power plant, the blade moment angle (mean value and standard deviation) of the blade 3 of the wind motor, the rotating speed of a blade wheel of the wind motor, the wind speed and the torque load of the tower top are calculated, and the training sample comprises a data value corresponding to the sample parameter. In another embodiment, when the model generation method is applied to the application scenario of speech recognition, then the sample parameters thereof may include: the sample user, sample user's sample voice, sample user identification, etc., the training sample just includes: the speech recognition sample parameter corresponds to a data value.
In one embodiment, when the model generation method is applied to an application scenario of beam bridge structure safety index prediction, then the sample parameters may include: the method comprises the following steps of beam strain (mean value and standard deviation), beam vertical displacement (mean value and standard deviation), beam cracking risk level, beam prestress loss risk level and load risk level, wherein a training sample comprises a data value corresponding to the sample parameter.
In one embodiment, when the method of model generation is applied to the application scenario of wind turbine generator fault prediction, then the sample parameters may include: the wind speed, the rotating speed set value, the rotating speed of the generator, the active power, the temperature in the engine room, the vibration of a transmission chain, the angle of a blade, the temperature of a front bearing of the generator and the temperature of a rear bearing of the generator, and the training sample comprises a data value corresponding to the parameter of the sample.
The embodiment of the present application further provides an apparatus 200 for generating a model, as shown in fig. 2, the apparatus 200 includes:
a first obtaining module 210, configured to obtain sample parameters of a neural network model to be created;
the second obtaining module 220 is configured to obtain a training data sample of the neural network model according to the sample parameter;
the creating module 230 is configured to receive a neural network creating instruction input by a user, create a neural network model, where the neural network creating instruction includes types of network layers included in the neural network model, an activation state of each network layer, a number of included neurons, and an input data dimension;
and the training module 240 is configured to train the neural network model according to the training data sample, the training method, and the training sample strategy input by the user, so as to obtain a target neural network model.
As an alternative embodiment, the apparatus 200 further comprises:
the third acquisition module is used for acquiring a model convergence curve of the target neural network model and judging the convergence of the target neural network model according to the loss value and the metric value in the model convergence curve;
the first judgment module is used for judging the convergence of the target neural network model if the loss value is smaller than a preset loss threshold value and the metric value is smaller than a preset metric threshold value;
and the second judging module is used for judging that the target neural network model does not converge if the loss value is greater than or equal to the preset loss threshold or the metric value is greater than or equal to the preset metric threshold, and prompting the user to modify the compiling parameters in the training sample strategy and the training method again.
As an alternative embodiment, the apparatus 200 further comprises:
the scoring module is used for obtaining a verification data set according to the verification splitting parameters in the training sample strategy and carrying out verification scoring processing on the target neural network model according to the verification data set to obtain a verification score;
the third judgment module is used for re-executing the training sample strategy according to the training data sample, the training method and the user input if the verification score is lower than the preset score threshold value, and training the neural network model to obtain a target neural network model;
and the fourth judging module is used for deriving the target neural network model if the verification score is equal to or higher than a preset score threshold value.
As an optional implementation manner, the second obtaining module 220 is specifically configured to obtain an import data sample according to a sample parameter; and according to a preset data processing method, carrying out extraction-conversion-loading processing on the imported data sample to obtain a training data sample.
The embodiment of the application provides a model generation device 200, wherein the device 200 is applied to computer equipment, and the computer equipment acquires sample parameters of a neural network model to be created; acquiring a training data sample of the neural network model according to the sample parameters; then, the computer equipment receives a neural network establishing instruction input by a user and establishes a neural network model, wherein the neural network establishing instruction comprises the type of each network layer contained in the neural network model, the activation state of each network layer, the number of contained neurons and the dimension of input data; and finally, the computer equipment trains the neural network model according to the training data sample, the training method and the training sample strategy input by the user to obtain the target neural network model. By adopting the method, the generation of the neural network model can be realized in a man-machine interaction mode.
It should be understood that although the steps in the flowcharts of fig. 1 and 2 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 1 and 2 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performing the sub-steps or stages is not necessarily sequential, but may be performed alternately or alternately with other steps or at least some of the sub-steps or stages of other steps.
For specific definition of the model generation device, reference may be made to the above definition of the model generation method, which is not described herein again. The modules in the model generation apparatus can be implemented in whole or in part by software, hardware, and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 3. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a method of model generation. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the architecture shown in fig. 3 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
It will be understood by those of ordinary skill in the art that all or a portion of the processes of the methods of the embodiments described above may be implemented by a computer program that may be stored on a non-volatile computer-readable storage medium, which when executed, may include the processes of the embodiments of the methods described above, wherein any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method of model generation, the method comprising:
acquiring sample parameters of a neural network model to be created;
acquiring a training data sample of the neural network model according to the sample parameter;
receiving a neural network creating instruction input by a user, creating the neural network model, wherein the neural network creating instruction comprises the types of all network layers included in the neural network model, the activation state of each network layer, the number of included neurons and an input data dimension;
and training the neural network model according to the training data sample, the training method and a training sample strategy input by a user to obtain a target neural network model.
2. The method of claim 1, further comprising:
acquiring a model convergence curve of the target neural network model, and judging the convergence of the target neural network model according to a loss value and a metric value in the model convergence curve;
if the loss value is less than the preset loss threshold and the metric value is less than the preset metric threshold, determining that the target neural network model converges;
and if the loss value is greater than or equal to the preset loss threshold value or the metric value is greater than or equal to the preset metric threshold value, judging that the target neural network model does not converge, and prompting a user to modify the training sample strategy and the compiling parameters in the training method again.
3. The method of claim 2, further comprising:
obtaining a verification data set according to a verification splitting parameter in the training sample strategy, and performing verification scoring processing on the target neural network model according to the verification data set to obtain a verification score;
if the verification score is lower than a preset score threshold value, the step of training the neural network model according to the training data sample, the training method and a training sample strategy input by a user is executed again to obtain a target neural network model;
deriving the target neural network model if the verification score is equal to or above the preset score threshold.
4. The method of claim 1, wherein obtaining training data samples of the neural network model based on the sample parameters comprises:
acquiring an imported data sample according to the sample parameter;
and according to a preset data processing method, carrying out extraction-conversion-loading processing on the imported data sample to obtain the training data sample.
5. An apparatus for model generation, the apparatus comprising:
the first acquisition module is used for acquiring sample parameters of a neural network model to be created;
the second acquisition module is used for acquiring a training data sample of the neural network model according to the sample parameter;
the creating module is used for receiving a neural network creating instruction input by a user and creating the neural network model, wherein the neural network creating instruction comprises the types of all network layers contained in the neural network model, the activation state of each network layer, the number of contained neurons and the dimension of input data;
and the training module is used for training the neural network model according to the training data sample, the training method and the training sample strategy input by the user to obtain the target neural network model.
6. The apparatus of claim 5, further comprising:
the third acquisition module is used for acquiring a model convergence curve of the target neural network model and judging the convergence of the target neural network model according to the loss value and the metric value in the model convergence curve;
a first determining module, configured to determine that the target neural network model converges if the loss value is less than the preset loss threshold and the metric value is less than the preset metric threshold;
and the second judging module is used for judging that the target neural network model does not converge if the loss value is greater than or equal to the preset loss threshold or the metric value is greater than or equal to the preset metric threshold, and prompting a user to modify the training sample strategy and the compiling parameters in the training method again.
7. The apparatus of claim 5, further comprising:
the scoring module is used for obtaining a verification data set according to the verification splitting parameters in the training sample strategy and carrying out verification scoring processing on the target neural network model according to the verification data set to obtain a verification score;
a third judging module, configured to, if the verification score is lower than a preset score threshold, re-execute the training sample strategy according to the training data sample, the training method, and the user input, and train the neural network model to obtain a target neural network model;
a fourth determining module for deriving the target neural network model if the verification score is equal to or higher than the preset score threshold.
8. The apparatus according to claim 5, wherein the second obtaining module is specifically configured to obtain an import data sample according to the sample parameter;
and according to a preset data processing method, carrying out extraction-conversion-loading processing on the imported data sample to obtain the training data sample.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 4 when executing the computer program.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010196133.6A CN111428869A (en) | 2020-03-19 | 2020-03-19 | Model generation method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010196133.6A CN111428869A (en) | 2020-03-19 | 2020-03-19 | Model generation method and device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111428869A true CN111428869A (en) | 2020-07-17 |
Family
ID=71548505
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010196133.6A Pending CN111428869A (en) | 2020-03-19 | 2020-03-19 | Model generation method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111428869A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112884052A (en) * | 2021-02-26 | 2021-06-01 | 北京激浊扬清文化科技有限公司 | Method and device for extracting structural modal parameters, computer equipment and storage medium |
| CN113313575A (en) * | 2021-06-08 | 2021-08-27 | 支付宝(杭州)信息技术有限公司 | Method and device for determining risk identification model |
| WO2024067143A1 (en) * | 2022-09-26 | 2024-04-04 | 华为技术有限公司 | Information transmission method, apparatus and system |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107766940A (en) * | 2017-11-20 | 2018-03-06 | 北京百度网讯科技有限公司 | Method and apparatus for generation model |
-
2020
- 2020-03-19 CN CN202010196133.6A patent/CN111428869A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107766940A (en) * | 2017-11-20 | 2018-03-06 | 北京百度网讯科技有限公司 | Method and apparatus for generation model |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112884052A (en) * | 2021-02-26 | 2021-06-01 | 北京激浊扬清文化科技有限公司 | Method and device for extracting structural modal parameters, computer equipment and storage medium |
| CN113313575A (en) * | 2021-06-08 | 2021-08-27 | 支付宝(杭州)信息技术有限公司 | Method and device for determining risk identification model |
| WO2024067143A1 (en) * | 2022-09-26 | 2024-04-04 | 华为技术有限公司 | Information transmission method, apparatus and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102422729B1 (en) | Learning Data Augmentation Policy | |
| CN111428869A (en) | Model generation method and device, computer equipment and storage medium | |
| CN110688288B (en) | Automatic test method, device, equipment and storage medium based on artificial intelligence | |
| Sola et al. | Importance of input data normalization for the application of neural networks to complex industrial problems | |
| KR20190028531A (en) | Training machine learning models for multiple machine learning tasks | |
| CN114048903A (en) | Intelligent optimization method for power grid safe operation strategy based on deep reinforcement learning | |
| US20230027527A1 (en) | Identity authentication method, and method and apparatus for training identity authentication model | |
| CN111598213A (en) | Network training method, data identification method, device, equipment and medium | |
| CN111160049B (en) | Text translation method, apparatus, machine translation system, and storage medium | |
| CN114186609A (en) | Model training method and device | |
| CN113627087B (en) | Production system parameter self-adaptive optimization method based on logistics simulation and machine learning | |
| CN113236487B (en) | Wind power plant noise control method, system, device and readable storage medium | |
| CN111142378A (en) | Neural network optimization method of biped robot neural network controller | |
| CN107665315B (en) | Role and trust-based access control method suitable for Hadoop | |
| US20220129248A1 (en) | Apparatus and method for generating random numbers based on reinforcement learning | |
| CN111245008B (en) | Wind field cooperative control method and device | |
| CN110457393B (en) | Information sharing method and related product | |
| CN113033422A (en) | Face detection method, system, equipment and storage medium based on edge calculation | |
| CN112115509A (en) | Data generation method and device | |
| CN113945974B (en) | Seismic data reconstruction method, device, equipment and medium | |
| CN111033532B (en) | Training method and system for generating countermeasure network, electronic device and storage medium | |
| CN117648946A (en) | DNN model automatic generation method for safety key system | |
| CN117975475A (en) | Visual language model instruction fine tuning method and device | |
| Martínez-Cruz et al. | Functional Verification of Digital Systems Using Meta-Heuristic Algorithms | |
| CN117077767A (en) | AIGC model establishing method based on federal learning and architecture searching |
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 |