CN115456162B - Training method and device for user behavior habit learning model based on virtual interaction - Google Patents
Training method and device for user behavior habit learning model based on virtual interaction Download PDFInfo
- Publication number
- CN115456162B CN115456162B CN202210968814.9A CN202210968814A CN115456162B CN 115456162 B CN115456162 B CN 115456162B CN 202210968814 A CN202210968814 A CN 202210968814A CN 115456162 B CN115456162 B CN 115456162B
- Authority
- CN
- China
- Prior art keywords
- virtual
- behavior
- learning model
- virtual person
- user
- 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
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/90335—Query processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T19/00—Manipulating 3D models or images for computer graphics
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Computer Hardware Design (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Graphics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention discloses a training method of a user behavior habit learning model based on virtual interaction, which comprises the following steps: setting a walking path of each virtual person in the virtual scene; controlling a virtual person to act according to a set walking path in a virtual scene within a preset time period, and collecting behavior data uploaded by each intelligent household device triggered when the virtual person walks according to the walking path; taking the collected behavior data as training data of the corresponding virtual person, and taking a walking path corresponding to the virtual person as a corresponding behavior habit marking; training the user behavior habit learning model based on the training data and the behavior habit labels corresponding to each virtual person to obtain a trained user behavior habit learning model. The method can save time and labor to construct training data, and greatly improves the model training efficiency. In addition, a training device, equipment and storage medium of the intelligent home control model based on virtual interaction are also provided.
Description
Technical Field
The present invention relates to the field of data processing technologies, and in particular, to a training method, apparatus, device, and storage medium for a learning model of user behavior habit based on virtual interaction.
Background
Along with the development of the internet of things technology, intelligent home starts to walk into life of people. However, a general problem in the industry is that smart home is not smart enough, and almost all controls still require users to actively operate through interaction modes such as key and voice. Although indeed more convenient than traditional home mode, still not reach the intelligent requirement of whole house. The active participation degree of the user required by the current intelligent home product when executing actions and scenes is too high, and a large amount of personal resources of the user are occupied. In order to achieve intellectualization, some scenes or operation modes can be preset to achieve intelligent control in the prior art, however, the operation is complicated when the scene or operation mode is preset, the preset scene or operation mode can not truly meet the actual requirements of users in many cases, and the deviation from the habit of the users is large, so that the preset scene or operation mode cannot be used.
Along with development of machine learning technology, at present, people propose to automatically learn the behavior habit of a user through machine learning, and the purpose of intelligently controlling intelligent household equipment is achieved according to the behavior habit of the user, for example, if the user has the habit of opening indoor light after entering a door, through machine learning, when the user enters the door, the automatic indoor light opening can be realized, and intelligent control can be realized without active operation of the user. However, since machine learning requires a large amount of training data, the acquisition of the training data often requires a long time, and manual labeling is required for the training data, not only is time-consuming and labor-consuming, but also the acquired training data often is various, and a series of normalization processes are required to be performed, so that the model training difficulty is increased, which is one of the reasons that full house intelligence is not fully popularized at present.
Disclosure of Invention
Based on the above, it is necessary to provide a training method, device, equipment and storage medium for a user behavior habit learning model based on virtual interaction, aiming at the problems of time and effort consumption and great model training difficulty in acquiring training data. The method realizes the quick and simple acquisition of the training data with the unified data format, and greatly reduces the difficulty of model training.
A training method of a user behavior habit learning model based on virtual interactions, the method comprising:
setting a walking path of each virtual person in a virtual scene, wherein each virtual person corresponds to a unique virtual person identifier, the walking path is formed by a series of actions, and each action corresponds to a corresponding time point;
controlling the virtual person to act according to the set walking path in the virtual scene within a preset time period, and simultaneously collecting behavior data uploaded by each intelligent home triggered when the virtual person walks according to the walking path in the virtual scene, wherein the behavior data comprises: action and corresponding time points;
taking the behavior data corresponding to each virtual person identifier acquired in the preset time period as training data of the corresponding virtual person, and taking the walking path corresponding to the virtual person as a corresponding behavior habit mark;
training the user behavior habit learning model based on the training data and the behavior habit labels corresponding to each virtual person to obtain a trained user behavior habit learning model.
A training device of an intelligent home control model based on virtual interactions, the device comprising:
the system comprises a setting module, a processing module and a processing module, wherein the setting module is used for setting a walking path of each virtual person in a virtual scene, each virtual person corresponds to a unique virtual person identifier, the walking path is formed by a series of actions, and each action corresponds to a corresponding time point;
the acquisition module is used for controlling the virtual person to act according to the set walking path in the virtual scene within a preset time period, and simultaneously acquiring behavior data uploaded by each intelligent home triggered when the virtual person walks according to the walking path in the virtual scene, wherein the behavior data comprises: action and corresponding time points;
the building module is used for taking the behavior data corresponding to each virtual person identifier acquired in the preset time period as training data of the corresponding virtual person and taking the walking path corresponding to the virtual person as corresponding behavior habit marking;
the training module is used for training the user behavior habit learning model based on the training data and the behavior habit labels corresponding to each virtual person, and obtaining a trained user behavior habit learning model.
A computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of:
setting a walking path of each virtual person in a virtual scene, wherein each virtual person corresponds to a unique virtual person identifier, the walking path is formed by a series of actions, and each action corresponds to a corresponding time point;
controlling the virtual person to act according to the set walking path in the virtual scene within a preset time period, and simultaneously collecting behavior data uploaded by each intelligent home triggered when the virtual person walks according to the walking path in the virtual scene, wherein the behavior data comprises: action and corresponding time points;
taking the behavior data corresponding to each virtual person identifier acquired in the preset time period as training data of the corresponding virtual person, and taking the walking path corresponding to the virtual person as a corresponding behavior habit mark;
training the user behavior habit learning model based on the training data and the behavior habit labels corresponding to each virtual person to obtain a trained user behavior habit learning model.
A computer-readable storage medium, comprising: a computer program is stored which, when executed by a processor, causes the processor to perform the steps of:
setting a walking path of each virtual person in a virtual scene, wherein each virtual person corresponds to a unique virtual person identifier, the walking path is formed by a series of actions, and each action corresponds to a corresponding time point;
controlling the virtual person to act according to the set walking path in the virtual scene within a preset time period, and simultaneously collecting behavior data uploaded by each intelligent home triggered when the virtual person walks according to the walking path in the virtual scene, wherein the behavior data comprises: action and corresponding time points;
taking the behavior data corresponding to each virtual person identifier acquired in the preset time period as training data of the corresponding virtual person, and taking the walking path corresponding to the virtual person as a corresponding behavior habit mark;
training the user behavior habit learning model based on the training data and the behavior habit labels corresponding to each virtual person to obtain a trained user behavior habit learning model.
According to the training method, the device, the computer equipment and the storage medium for the virtual interaction-based user behavior habit learning model, the behavior operation habits of a real person in a real environment are simulated by setting the walking path of the virtual person in the virtual scene, when the virtual person operates the intelligent home equipment in the virtual scene, the corresponding virtual intelligent home equipment is triggered like the real environment, the virtual intelligent home equipment automatically uploads the behavior data of the virtual person to the cloud, so that the behavior data corresponding to each virtual person in a preset time period can be obtained from the cloud as training data, and meanwhile, the preset virtual pedestrian walking path can be used as the corresponding behavior habit marking, so that training data can be quickly obtained, the training data is automatically marked, time and labor are saved, the obtained training data are in a unified format, the processing difficulty is greatly reduced, the user behavior habit learning model is trained based on the training data and the corresponding behavior habit marking, and the user behavior habit learning model which can be used for learning the user behavior habits is obtained.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Wherein:
FIG. 1 is a flow diagram of a method of training a user behavior habit learning model based on virtual interactions in one embodiment;
FIG. 2 is a block diagram of a training device based on a virtual interaction user behavior habit learning model in one embodiment;
FIG. 3 is a schematic diagram of the internal structure of a computer device in one embodiment.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, in one embodiment, a training method of a user behavior habit learning model based on virtual interaction includes:
The virtual person can be a 3D virtual person or a simple virtual image. The virtual scene is a virtual scene which can be subjected to real interaction and is constructed through VR technology. Various virtual intelligent home devices are arranged in the virtual scene, the corresponding intelligent home devices can be controlled to be opened or closed in the virtual scene by operating the intelligent home devices in the virtual scene, and the linkage between the virtual scene and the real devices can be realized, namely, the intelligent home devices in the real scene are controlled by a switch in the virtual scene. In order to simply acquire training data for model training, a walking path of a virtual person in a virtual scene is firstly set, wherein the walking path is formed by a series of actions, for example, an intelligent lock of a family door is firstly opened, then light is opened, then the person walks to the front of a refrigerator, then the refrigerator is opened, then the person turns on the refrigerator, then the person walks to a sofa, then the person sits down, and then a television and other series of actions are opened. Each action is provided with a corresponding time point, and then the corresponding action is executed according to the specific time point, namely, the virtual person simulates the behavior habit of the real person.
In order to enable the trained user behavior habit learning model to be used for subsequent learning of the behavior habits of the real user. The virtual person adopts a 3D virtual person, so that the behavior habit of the real person can be simulated to the greatest extent, and the behavior habit of the real person can be learned to the greatest extent. The 3D virtual person is set because the 3D virtual person is required to imitate the actions of the person as much as possible, so that the camera can predict the actions of the shot virtual person, for example, the refrigerator is opened when the virtual person is identified to take the things on hand each time and walks in front of the refrigerator, after the behavior habits of the virtual persons are learned, the refrigerator is automatically opened when the virtual person is identified to take the things on hand next time and walks in front of the refrigerator, and the model obtained based on the training is used for better subsequent identification of the behavior habits of the real person.
The preset period of time refers to a set period of time, for example, one week. Since the behavior habit of the virtual person (simulating the real person) is to be learned, behavior habit data of the virtual person within a period of time needs to be collected and uploaded through the smart home, for example, when a user opens the smart refrigerator, the smart refrigerator can upload the action of opening the refrigerator to the cloud record at what time point.
And 106, taking the behavior data corresponding to each virtual person identifier acquired in a preset time period as training data of the corresponding virtual person, and taking the walking path corresponding to the virtual person as a corresponding behavior habit mark.
A plurality of virtual persons are preset, each virtual person is provided with a corresponding virtual person identifier (such as faceID), the virtual person identifier is associated with corresponding behavior data, the behavior data are used as training data, and meanwhile, the corresponding walking path which is set before is used as behavior habit marking, so that the problems of difficult data collection, long period and difficult marking in the prior art are solved, and the method is equivalent to a traditional data set construction mode.
And step 108, training the user behavior habit learning model based on the training data and the behavior habit labels corresponding to each virtual person to obtain a trained user behavior habit learning model.
The initial user behavior habit model is built by taking a neural network model as a basis, a trained user behavior habit learning model can be obtained through learning based on training data, the trained user behavior habit learning model can be used for learning of user behavior habits, and then when the user behavior habits need to be learned, the user behavior habits can be predicted only by taking the user behavior data collected in a preset time period as the input of the model, and then corresponding intelligent household equipment can be controlled based on the user behavior habits.
According to the training method of the user behavior habit learning model based on virtual interaction, the behavior operation habits of a real person in a real environment are simulated by setting the walking path of the virtual person in the virtual scene, when the virtual person operates the intelligent home equipment in the virtual scene, the corresponding virtual intelligent home equipment is triggered like the real environment, the virtual intelligent home equipment automatically uploads the behavior data of the virtual person to the cloud, so that the behavior data corresponding to each virtual person in a preset time period can be obtained from the cloud as training data, and meanwhile, the preset virtual pedestrian walking path can be used as the corresponding behavior habit marking, so that training data can be quickly obtained, the training data can be automatically marked, time and labor are saved, the obtained training data are in a unified format, the processing difficulty is greatly reduced, the user behavior habit learning model is trained based on the training data and the corresponding behavior habit marking, and the user behavior habit learning model which can be used for learning the behavior habits of the user is obtained.
In one embodiment, the method further comprises:
taking the virtual person identification and training data corresponding to the virtual person identification as input of the trained user behavior habit learning model, wherein the trained user behavior habit learning model obtains a virtual person behavior habit learning model corresponding to the virtual person identification through learning of the training data;
when the virtual person identification and the current action are obtained, the next action of the virtual person is output by adopting the virtual person behavior habit learning model according to the current action, and intelligent control is performed on the intelligent household equipment based on the next action.
The training method comprises the steps of training a behavior habit learning model of a virtual person, learning the behavior habit of the virtual person by using the trained user behavior habit learning model, obtaining a virtual person behavior habit learning model corresponding to the virtual person, and predicting the next action of the virtual person according to the virtual person behavior habit learning model, so that intelligent control is performed on intelligent household equipment based on the next action. The current action corresponds to the current time, and the current action and the current time point are used as corresponding virtual person behavior habit learning models to predict the next action of the virtual person. The trained user behavior habit learning model is a general basic model obtained through learning, and in order to predict the behavior habit of a specific virtual person, a corresponding virtual person behavior habit learning model needs to be obtained through learning based on the basic model. For example, the behavior habit of the dummy is learned to turn on the door first and then turn on the light. When the action of opening the door by the dummy is detected, the indoor light is automatically turned on.
In one embodiment, the method further comprises:
when the real user does not have corresponding historical behavior data yet, acquiring behavior habits of the real user in a questionnaire mode, performing similarity calculation based on the behavior habits of the real user and the behavior habits of N virtual persons in a database, and taking a virtual person behavior habit learning model corresponding to a virtual person identifier with the highest similarity as an initial behavior habit learning model of the real user;
when the use behavior data of the real user uploaded by the intelligent home equipment in the real environment is obtained within a period of time, updating the initial behavior habit learning model based on the use behavior data until the personalized behavior habit learning model corresponding to the real user is obtained.
At present, users can learn the real behavior habit of the users in a long time after buying the intelligent home products and returning home, and then the full-house intelligent control can be realized, which is very unfriendly to the users who purchase the intelligent home equipment for the first time, namely, the users can not enjoy the effect of the full-house intelligent control in a short time. Therefore, in the scheme, a plurality of (N) virtual people behavior habit learning models are obtained through training, and the N virtual people represent behavior habits of a real user respectively. In order to enable a customer to purchase intelligent home equipment for enjoying the effect of intelligent control, the behavior habit of a real user is obtained through a questionnaire mode, then similarity calculation is carried out on the behavior habit and the behavior habits of N virtual persons, the initial behavior habit learning model with the highest similarity is used as an initial behavior habit learning model of the real user, specifically, the virtual person with the highest similarity is downloaded into a real intelligent home scene of the user, and the model is quite similar to the user, so that the user can enjoy the effect of intelligent control of a whole house quickly, certainly has difference with the behavior habit of the real user, and the model can be automatically updated when the using behavior data of the real user in a real environment are collected later, so that the personalized behavior habit learning model corresponding to the real user is obtained. And because the previous behavior habits are very similar, namely the initialization is very close, the time spent for subsequently updating the model is greatly shortened, and the learning efficiency of the personalized behavior habit learning model is improved.
In one embodiment, the similarity calculation based on the behavior habit of the real user and the behavior habits of the N virtual persons in the database includes: performing action association splitting on the behavior habits of the real user, taking every two associated actions in the behavior habits as a target unit, and determining a plurality of target units corresponding to the real user; obtaining a plurality of standard units corresponding to each virtual person; and respectively matching a plurality of target units of the real user with a plurality of standard units of each virtual person to obtain the similarity between the real user and each virtual person.
In order to calculate the similarity, the behavior habits of the real user are subjected to action association and separation, and two adjacent association actions are used as one target unit, for example, the user is used to turn on a lamp after entering a door, then turns on a refrigerator, thus the door-on lamp is used as one target unit, and the refrigerator is turned on as one target unit, so that a plurality of target units are obtained. Then, similarly, the behavior habit of each virtual person is subjected to action association splitting, and for distinguishing, two associated actions of the virtual person are called as standard units. By matching a plurality of target units of the real user with a plurality of standard units of the virtual person, the more the matching quantity is, the higher the similarity is.
In one embodiment, the method further comprises: when the current action of the real user is obtained, taking the user identification and the current action of the real user as the input of a corresponding personalized behavior habit learning model, and outputting the next walking habit action of the real user; and sending a control instruction to the intelligent household equipment based on the habit actions of the next walking, wherein the control instruction is used for controlling the corresponding intelligent household equipment to be opened or closed.
After training to obtain the personalized behavior habit learning model corresponding to the real user, the personalized behavior habit learning model can be utilized to conduct next prediction of the user habit, so that control instructions can be sent in advance, the behavior of the client can be prejudged in advance, and therefore more intelligent service is provided for the client.
In one embodiment, the method further comprises: when a set of intelligent household equipment is shared by multiple people in a real scene, each person wears an induction bracelet, and the intelligent household equipment identifies a user identifier through the induction bracelet, and uploads the user identifier and corresponding operation actions to the cloud for storage.
In order to learn the behavior habits of each person, the users need to be distinguished and identified, i.e. different users can be identified. The aim of identifying the user is achieved by providing each member with an intelligent bracelet, so that the user identification and the corresponding actions for operating the intelligent home can be uploaded to the cloud for storage, and the personalized behavior habit learning model corresponding to each family member can be trained and obtained.
In one embodiment, the training the learning model of the user behavior habit based on the training data and the behavior habit labels corresponding to each virtual person to obtain a trained learning model of the user behavior habit includes: determining a training data set and a testing data set according to the obtained training data and behavior habit labels corresponding to each virtual person; training the user behavior habit learning model based on the training data set to obtain a target user behavior habit learning model; and verifying the target user behavior habit learning model based on the test data set, and taking the target user behavior habit learning model as the trained user behavior habit learning model if the verification meets the requirements.
In order to verify the performance of the model obtained through training, during model training, training data are divided into a training data set and a test data set, the test data set verifies the model obtained through training, and the model is judged to be training completion only after verification is qualified. The training data collected in the virtual scene can be used for training the model and completing model verification, so that the training data of the virtual scene can be used for training and completing the user behavior habit learning model.
As shown in fig. 2, a training device of an intelligent home control model based on virtual interaction is provided, where the device includes:
the setting module 202 is configured to set a walking path of each virtual person in the virtual scene, where each virtual person corresponds to a unique virtual person identifier, the walking path is formed by a series of actions, and each action corresponds to a corresponding time point;
the collection module 204 is configured to control the virtual person to perform an action according to the set walking path in the virtual scene within a preset time period, and collect behavior data uploaded by each smart home triggered when the virtual person walks according to the walking path in the virtual scene, where the behavior data includes: action and corresponding time points;
the construction module 206 is configured to take the behavior data corresponding to each virtual person identifier collected in the preset time period as training data of a corresponding virtual person, and take a walking path corresponding to the virtual person as a corresponding behavior habit label;
the training module 208 is configured to train the user behavior habit learning model based on the training data and the behavior habit labels corresponding to each dummy, and obtain a trained user behavior habit learning model.
In one embodiment, the apparatus further comprises:
the learning module is used for taking the virtual person identification and training data corresponding to the virtual person identification as input of the trained user behavior habit learning model, and the trained user behavior habit learning model obtains a virtual person behavior habit learning model corresponding to the virtual person identification through learning of the training data;
the first control module is used for outputting the next action of the virtual person by adopting the virtual person behavior habit learning model according to the current action when the virtual person identification and the current action are acquired, and intelligently controlling the intelligent household equipment based on the next action.
In one embodiment, the apparatus further comprises:
the computing module is used for acquiring the behavior habit of the real user in a questionnaire mode when the real user does not have the corresponding historical behavior data, carrying out similarity computation based on the behavior habit of the real user and the behavior habits of N virtual persons in a database, and taking a virtual person behavior habit learning model corresponding to the virtual person identification with the highest similarity as an initial behavior habit learning model of the real user;
and the updating module is used for updating the initial behavior habit learning model based on the usage behavior data when the usage behavior data of the real user uploaded by the intelligent household equipment in the real environment is acquired within a period of time until the personalized behavior habit learning model corresponding to the real user is obtained.
In one embodiment, the computing module is further configured to perform action association splitting on the behavior habits of the real user, and determine a plurality of target units corresponding to the real user by using the pairwise association actions in the behavior habits as a target unit; obtaining a plurality of standard units corresponding to each virtual person; and respectively matching a plurality of target units of the real user with a plurality of standard units of each virtual person to obtain the matching degree of the real user and each virtual person.
In one embodiment, the apparatus further comprises:
the output module is used for taking the user identification and the current action of the real user as the input of a corresponding personalized behavior habit learning model when the current action of the real user is obtained, and outputting the next walking habit action of the real user;
and the second control module is used for sending a control instruction to the intelligent household equipment based on the next walking habit action, wherein the control instruction is used for controlling the corresponding intelligent household equipment to be opened or closed.
In one embodiment, the apparatus further comprises:
and the uploading module is used for enabling each person to wear an induction bracelet when a set of intelligent household equipment is shared by multiple persons in the real scene, and the intelligent household equipment identifies the user identification through the induction bracelet and uploads the user identification and corresponding operation actions to the cloud for storage.
In one embodiment, the training module is further configured to determine a training data set and a test data set according to the obtained training data and behavior habit labels corresponding to each virtual person; training the user behavior habit learning model based on the training data set to obtain a target user behavior habit learning model; and verifying the target user behavior habit learning model based on the test data set, and taking the target user behavior habit learning model as the trained user behavior habit learning model if the verification meets the requirements.
FIG. 3 illustrates an internal block diagram of a computer device in one embodiment. The computer device may specifically be a terminal or a server. As shown in fig. 3, the computer device includes a processor, a memory, and a network interface connected by a system bus. The memory includes a nonvolatile storage medium and an internal memory. The non-volatile storage medium of the computer device has a storage operating system and may also have a computer program, which when executed by the processor, causes the processor to implement the training method of the user behavior habit learning model based on virtual interaction. The internal memory may also store a computer program that, when executed by the processor, causes the processor to perform the training method of the user behavior habit learning model based on virtual interactions described above. It will be appreciated by those skilled in the art that the structure shown in fig. 3 is merely a block diagram of a portion of the structure associated with the present application and does not constitute a limitation of the apparatus to which the present application is applied, and that a particular apparatus may include more or less components than those shown in the drawings, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having stored a computer program which, when executed by the processor, causes the processor to perform the steps of the above-described training method for a user behavior habit learning model based on virtual interactions.
In one embodiment, a computer readable storage medium is provided, storing a computer program, which when executed by a processor, causes the processor to perform the steps of the training method of the user behavior habit learning model based on virtual interactions.
It can be appreciated that the training method, the training device, the computer device and the computer readable storage medium for the user behavior habit learning model based on virtual interaction belong to a general inventive concept, and the embodiments are mutually applicable.
Those skilled in the art will appreciate that all or part of the processes in the methods of the above embodiments may be implemented by a computer program for instructing relevant hardware, where the program may be stored in a non-volatile computer readable storage medium, and where the program, when executed, may include processes in the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.
Claims (9)
1. A training method of a user behavior habit learning model based on virtual interaction, the method comprising:
setting a walking path of each virtual person in a virtual scene, wherein each virtual person corresponds to a unique virtual person identifier, the walking path is formed by a series of actions, and each action corresponds to a corresponding time point;
controlling the virtual person to act according to the set walking path in the virtual scene within a preset time period, and simultaneously collecting behavior data uploaded by each intelligent household device triggered when the virtual person walks according to the walking path in the virtual scene, wherein the behavior data comprises: action and corresponding time points;
taking the behavior data corresponding to each virtual person identifier acquired in the preset time period as training data of the corresponding virtual person, and taking the walking path corresponding to the virtual person as a corresponding behavior habit mark;
training the user behavior habit learning model based on training data and behavior habit labels corresponding to each virtual person to obtain a trained user behavior habit learning model;
the method further comprises the steps of:
when the real user does not have corresponding historical behavior data yet, acquiring behavior habits of the real user in a questionnaire mode, performing similarity calculation based on the behavior habits of the real user and the behavior habits of N virtual persons in a database, and taking a virtual person behavior habit learning model corresponding to a virtual person identifier with the highest similarity as an initial behavior habit learning model of the real user;
when the use behavior data of the real user uploaded by the intelligent home equipment in the real environment is obtained within a period of time, updating the initial behavior habit learning model based on the use behavior data until the personalized behavior habit learning model corresponding to the real user is obtained.
2. The method according to claim 1, wherein the method further comprises:
taking the virtual person identification and training data corresponding to the virtual person identification as input of the trained user behavior habit learning model, wherein the trained user behavior habit learning model obtains a virtual person behavior habit learning model corresponding to the virtual person identification through learning of the training data;
when the virtual person identification and the current action are obtained, the next action of the virtual person is output by adopting the virtual person behavior habit learning model according to the current action, and intelligent control is performed on the intelligent household equipment based on the next action.
3. The method according to claim 1, wherein the similarity calculation based on the behavior habits of the real user and the behavior habits of N virtual persons in a database comprises:
performing action association splitting on the behavior habits of the real user, taking every two associated actions in the behavior habits as a target unit, and determining a plurality of target units corresponding to the real user;
obtaining a plurality of standard units corresponding to each virtual person;
and respectively matching a plurality of target units of the real user with a plurality of standard units of each virtual person to obtain the matching degree of the real user and each virtual person.
4. The method according to claim 1, wherein the method further comprises:
when the current action of the real user is obtained, taking the user identification and the current action of the real user as the input of a corresponding personalized behavior habit learning model, and outputting the next walking habit action of the real user;
and sending a control instruction to the intelligent household equipment based on the habit actions of the next walking, wherein the control instruction is used for controlling the corresponding intelligent household equipment to be opened or closed.
5. The method according to claim 1, wherein the method further comprises:
when a set of intelligent household equipment is shared by multiple people in a real scene, each person wears an induction bracelet, and the intelligent household equipment identifies a user identifier through the induction bracelet, and uploads the user identifier and corresponding operation actions to the cloud for storage.
6. The method according to claim 1, wherein the training the user behavior habit learning model based on the training data and the behavior habit labels corresponding to each dummy to obtain a trained user behavior habit learning model includes:
determining a training data set and a testing data set according to the obtained training data and behavior habit labels corresponding to each virtual person;
training the user behavior habit learning model based on the training data set to obtain a target user behavior habit learning model;
and verifying the target user behavior habit learning model based on the test data set, and taking the target user behavior habit learning model as the trained user behavior habit learning model if the verification meets the requirements.
7. A training device for a user behavior habit learning model based on virtual interactions, the device comprising:
the system comprises a setting module, a processing module and a processing module, wherein the setting module is used for setting a walking path of each virtual person in a virtual scene, each virtual person corresponds to a unique virtual person identifier, the walking path is formed by a series of actions, and each action corresponds to a corresponding time point;
the acquisition module is used for controlling the virtual person to act according to the set walking path in the virtual scene within a preset time period, and simultaneously acquiring behavior data uploaded by each intelligent home triggered when the virtual person walks according to the walking path in the virtual scene, wherein the behavior data comprises: action and corresponding time points;
the building module is used for taking the behavior data corresponding to each virtual person identifier acquired in the preset time period as training data of the corresponding virtual person and taking the walking path corresponding to the virtual person as corresponding behavior habit marking;
the training module is used for training the user behavior habit learning model based on the training data and the behavior habit labels corresponding to each virtual person to obtain a trained user behavior habit learning model;
the computing module is used for acquiring the behavior habit of the real user in a questionnaire mode when the real user does not have the corresponding historical behavior data, carrying out similarity computation based on the behavior habit of the real user and the behavior habits of N virtual persons in a database, and taking a virtual person behavior habit learning model corresponding to the virtual person identification with the highest similarity as an initial behavior habit learning model of the real user;
and the updating module is used for updating the initial behavior habit learning model based on the usage behavior data when the usage behavior data of the real user uploaded by the intelligent household equipment in the real environment is acquired within a period of time until the personalized behavior habit learning model corresponding to the real user is obtained.
8. A computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the training method of the virtual interaction based user behavior habit learning model of any one of claims 1 to 6.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program which, when executed by the processor, causes the processor to perform the steps of the training method of the virtual interaction-based user behavior habit learning model of any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210968814.9A CN115456162B (en) | 2022-08-12 | 2022-08-12 | Training method and device for user behavior habit learning model based on virtual interaction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210968814.9A CN115456162B (en) | 2022-08-12 | 2022-08-12 | Training method and device for user behavior habit learning model based on virtual interaction |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115456162A CN115456162A (en) | 2022-12-09 |
CN115456162B true CN115456162B (en) | 2023-07-14 |
Family
ID=84298164
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210968814.9A Active CN115456162B (en) | 2022-08-12 | 2022-08-12 | Training method and device for user behavior habit learning model based on virtual interaction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115456162B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117273054B (en) * | 2023-09-28 | 2024-06-25 | 江苏八点八智能科技有限公司 | Virtual human interaction method and system applying different scenes |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112090062A (en) * | 2020-09-14 | 2020-12-18 | 成都拟合未来科技有限公司 | Interactive practice virtual scene display method, system, terminal and medium |
CN112162628A (en) * | 2020-09-01 | 2021-01-01 | 魔珐(上海)信息科技有限公司 | Multi-mode interaction method, device and system based on virtual role, storage medium and terminal |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11461689B2 (en) * | 2017-01-06 | 2022-10-04 | Sigurdur Runar Petursson | Techniques for automatically testing/learning the behavior of a system under test (SUT) |
WO2020088739A1 (en) * | 2018-10-29 | 2020-05-07 | Hexagon Technology Center Gmbh | Facility surveillance systems and methods |
CN109947029B (en) * | 2019-04-10 | 2022-05-10 | 广东工业大学 | Control method, device and equipment of intelligent household equipment |
CN114359658A (en) * | 2021-12-15 | 2022-04-15 | 深圳市优必选科技股份有限公司 | Training data generation method and device, terminal equipment and storage medium |
-
2022
- 2022-08-12 CN CN202210968814.9A patent/CN115456162B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112162628A (en) * | 2020-09-01 | 2021-01-01 | 魔珐(上海)信息科技有限公司 | Multi-mode interaction method, device and system based on virtual role, storage medium and terminal |
CN112090062A (en) * | 2020-09-14 | 2020-12-18 | 成都拟合未来科技有限公司 | Interactive practice virtual scene display method, system, terminal and medium |
Also Published As
Publication number | Publication date |
---|---|
CN115456162A (en) | 2022-12-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110533097B (en) | Image definition recognition method and device, electronic equipment and storage medium | |
CN109902222B (en) | Recommendation method and device | |
CN108905209B (en) | Method and system for detecting plug-in game, electronic equipment and storage medium | |
CN110166827B (en) | Video clip determination method and device, storage medium and electronic device | |
CN110339569B (en) | Method and device for controlling virtual role in game scene | |
CN108427708B (en) | Data processing method, data processing apparatus, storage medium, and electronic apparatus | |
CN108921221A (en) | Generation method, device, equipment and the storage medium of user characteristics | |
CN110249622A (en) | The camera exposure control of real-time Semantic Aware | |
CN109447164B (en) | A kind of motor behavior method for classifying modes, system and device | |
CN109784654A (en) | Task creating method, device, computer equipment and storage medium | |
CN115456162B (en) | Training method and device for user behavior habit learning model based on virtual interaction | |
CN109816758B (en) | Two-dimensional character animation generation method and device based on neural network | |
CN115063874B (en) | Control method, device and equipment of intelligent household equipment and storage medium | |
CN112791414B (en) | Plug-in recognition model training method and device, electronic equipment and storage medium | |
CN110750523A (en) | Data annotation method, system, computer equipment and storage medium | |
CN112905876A (en) | Information pushing method and device based on deep learning and computer equipment | |
CN112116589B (en) | Method, device, equipment and computer readable storage medium for evaluating virtual image | |
CN114064974A (en) | Information processing method, information processing apparatus, electronic device, storage medium, and program product | |
CN112817563A (en) | Target attribute configuration information determination method, computer device, and storage medium | |
CN109670623A (en) | Neural net prediction method and device | |
CN112559777A (en) | Content item delivery method and device, computer equipment and storage medium | |
Rahtz et al. | Safe deep rl in 3d environments using human feedback | |
CN117036834B (en) | Data classification method and device based on artificial intelligence and electronic equipment | |
CN113761355A (en) | Information recommendation method, device, equipment and computer readable storage medium | |
CN111089388A (en) | Method and system for controlling air conditioner, air conditioner and household appliance |
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 |