WO2024045128A1 - Procédé et appareil d'affichage de modèle d'intelligence artificielle, dispositif électronique et support de stockage - Google Patents

Procédé et appareil d'affichage de modèle d'intelligence artificielle, dispositif électronique et support de stockage Download PDF

Info

Publication number
WO2024045128A1
WO2024045128A1 PCT/CN2022/116490 CN2022116490W WO2024045128A1 WO 2024045128 A1 WO2024045128 A1 WO 2024045128A1 CN 2022116490 W CN2022116490 W CN 2022116490W WO 2024045128 A1 WO2024045128 A1 WO 2024045128A1
Authority
WO
WIPO (PCT)
Prior art keywords
model
display
artificial intelligence
display area
data set
Prior art date
Application number
PCT/CN2022/116490
Other languages
English (en)
Chinese (zh)
Inventor
于禾
王琪
刘展宏
田德钰
Original Assignee
西门子股份公司
西门子（中国）有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 西门子股份公司, 西门子（中国）有限公司 filed Critical 西门子股份公司
Priority to PCT/CN2022/116490 priority Critical patent/WO2024045128A1/fr
Publication of WO2024045128A1 publication Critical patent/WO2024045128A1/fr

Links

Images

Classifications

    • 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
    • G06F16/901—Indexing; Data structures therefor; Storage structures
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology

Definitions

  • the present invention relates to the technical field of artificial intelligence (Artificial Intelligence, AI), in particular to methods, devices, electronic equipment and storage media for displaying AI models.
  • AI Artificial Intelligence
  • MLOps Machine Learning Operations
  • DevOps development and operations
  • the embodiment of the present invention provides an AI model display method, device, electronic equipment and storage medium.
  • a method for displaying an AI model includes:
  • the graphical file is displayed in the first display area.
  • the simple information of the AI model is no longer displayed, but the hierarchical system among the parameters of the AI model is displayed graphically, which improves the display efficiency of the model and facilitates users to deeply understand the model performance.
  • it also includes:
  • the text description information and/or the graphic description information are displayed in a second display area surrounding the first display area.
  • the data set is no longer separated from the AI model.
  • the display efficiency of the model is improved, and it is also convenient for users to have a deep understanding of the model performance.
  • it also includes:
  • model code which includes code snippets and comments for the code snippets
  • the currently running code fragment and its annotations are displayed in the third display area, and the current running data provided to the currently running code fragment from the selected data set is synchronously displayed in a graphical manner in the fourth display area.
  • it also includes:
  • the monitoring variable value is displayed in the fifth display area.
  • it also includes:
  • the graphical file is displayed in the sixth display area.
  • An AI model display device the device includes:
  • the receiving module is configured to receive model query requests containing model keywords
  • a determining module configured to determine the AI model corresponding to the model keyword based on the model query request
  • An acquisition module configured to acquire a graphical file associated with the AI model, wherein the graphical file contains a hierarchical system between predetermined parameters in the AI model, the hierarchical system is based on parsing the AI model Obtained from the model configuration file;
  • the display module is configured to display the graphical file in the first display area.
  • the simple information of the AI model is no longer displayed, but the hierarchical system among the parameters of the AI model is displayed graphically, which improves the display efficiency of the model and facilitates users to deeply understand the model performance.
  • the determination module is configured to determine a data set of the AI model, where the data set includes a training data set or a test database; determine text description information and graphical description information of the data set, The graphical description information is adapted to graphically display data in the data set;
  • the display module is configured to display the text description information and/or the graphical description information in a second display area surrounding the first display area.
  • the data set is no longer separated from the AI model.
  • the display efficiency of the model is improved, and it is also convenient for users to have a deep understanding of the model performance.
  • the acquisition module is configured to acquire the dependency library of the AI model; create a running environment of the AI model based on the dependency library; and deserialize the AI model to obtain Model code, the model code includes code snippets and comments for the code snippets; in the running environment, run the AI model based on the selected data set;
  • the display module is configured to display the currently running code snippets and their annotations in the third display area, and to graphically display the currently running code snippets and their annotations provided from the selected data set in a fourth display area.
  • the current running data of the code snippet is configured to display the currently running code snippets and their annotations in the third display area, and to graphically display the currently running code snippets and their annotations provided from the selected data set in a fourth display area.
  • the current running data of the code snippet is configured to display the currently running code snippets and their annotations in the third display area, and to graphically display the currently running code snippets and their annotations provided from the selected data set in a fourth display area.
  • the acquisition module is configured to determine a monitoring variable value when running the currently running code fragment based on the current running data
  • the display module is configured to display the monitored variable value in a fifth display area.
  • the acquisition module is configured to determine a model output result generated by running the currently running code fragment based on the current running data; generate the model output result graphical files;
  • the display module is configured to display the graphical file in the sixth display area.
  • An electronic device including:
  • the processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the AI model display method as described in any one of the above.
  • a computer-readable storage medium on which computer instructions are stored.
  • the method for displaying an AI model as described in any one of the above items is implemented.
  • a computer program product includes a computer program that, when executed by a processor, implements the AI model display method described in any one of the above items.
  • Figure 1 is a flow chart of a method for displaying an AI model according to an embodiment of the present invention.
  • Figure 2 is an exemplary schematic diagram illustrating the display effect of an AI model according to an embodiment of the present invention.
  • Figure 3 is an exemplary schematic diagram of a presentation process of an AI model according to an embodiment of the present invention.
  • Figure 4 is an exemplary schematic diagram of the subscription process and query process of the AI model according to the embodiment of the present invention.
  • Figure 5 is an exemplary schematic diagram of tracking model training in a code board and a vision board according to an embodiment of the present invention.
  • FIG. 6 is an exemplary structural diagram of an AI model display device according to an embodiment of the present invention.
  • FIG. 7 is an exemplary structural diagram of an electronic device according to an embodiment of the present invention.
  • Model training is performed on the developer side, while the management side only displays simple introduction information of the model. This simple introductory information is not enough to help users understand the value or performance of the model.
  • an AI model display method is proposed (which can be implemented as a plug-in or functional module, etc.), providing a vivid and scalable solution for MLOps services.
  • the implementation of the present invention makes model recognition more efficient, feedback and operation more friendly, and promotes automated and intelligent digital applications.
  • Figure 1 is a flow chart of a method for displaying an AI model according to an embodiment of the present invention. This method can be performed by a plug-in or functional module integrated into the MLOps service. As shown in Figure 1, the method 100 includes:
  • Step 101 Receive a model query request containing model keywords.
  • model keywords can include model name, model version number, model registrant identification, model registration time, model status, model life cycle stage, model duration, model user, etc.
  • Step 102 Based on the model query request, determine the AI model corresponding to the model keyword.
  • the model keyword is compared with the index of the stored AI model in the model library to determine the AI model corresponding to the model keyword.
  • the index can be implemented as the key of a key-value-pair in the model library.
  • the model query request can be sent to the model library, and the model keywords are compared with the index of the stored AI model in the model library.
  • the comparison process can use exact matching or simulated matching.
  • there are multiple model keywords. Comparing the model keywords with the index of the stored AI model includes: assigning a weight to each model keyword; fuzzy matching based on each model keyword and its weight. The method is compared with the index of the stored AI model. It can be seen that through weighted fuzzy matching, intelligent search AI models can be realized.
  • Step 103 Obtain a graphical file associated with the AI model, where the graphical file contains a hierarchical system between predetermined parameters in the AI model, and the hierarchical system is obtained based on parsing the model configuration file of the AI model.
  • a graphical file associated with the AI model is further obtained.
  • the graphical file contains a hierarchy between predetermined parameters in the AI model, where the hierarchy is obtained based on parsing the model configuration file of the AI model. For example, parse the model configuration file of the AI model to obtain the list of key parameters and the hierarchy between key parameters (for example, the next layer pointed to by a key parameter contains other key parameters).
  • a hierarchical system has a structure similar to a knowledge graph. You can graphically display several key parameters in the AI model in the graphical file, as well as graphically display the default values or introduction instructions of the key parameters, etc.
  • Step 104 Display the graphical file in the first display area.
  • the graphical file is displayed in the first display area.
  • the basic information of the model is further displayed in text in the first display area, such as model name, model version number, model registrant's identification, model registration time, model status, model life cycle stage, model duration, and model user ,etc.
  • model name e.g., model name, model version number, model registrant's identification, model registration time, model status, model life cycle stage, model duration, and model user ,etc.
  • the method 100 further includes: determining a data set (dataset) of the AI model, the data set including a training data set or a test database; determining text description information and graphical description information of the data set, wherein the graphical description information is adapted Display the data in the data set graphically; display text description information and/or graphic description information in a second display area surrounding the first display area. Therefore, the data set is not separated from the AI model. By displaying the description information of the data set, the display efficiency of the model is improved, and it is also convenient for users to have a deep understanding of the model performance.
  • the method 100 further includes: obtaining the dependency library of the AI model; creating a running environment for the artificial intelligence model based on the dependency library; deserializing the artificial intelligence model to obtain the model code, where the model code includes code fragments and comments for code snippets; in the running environment, run the artificial intelligence model based on the selected data set; display the currently running code snippets and their comments in the third display area, and graphically synchronize them in the fourth display area Displays the current run data from the selected data set provided to the currently running code snippet.
  • code snippet annotations are not only used to annotate code snippets, but can also be used to locate and track code snippets. It can be seen that by locating code snippets through annotations, the currently running code snippets and current running data are simultaneously displayed graphically, making it easier to understand the meaning behind the parameters and the model operation effect.
  • the method 100 further includes: determining a monitoring variable value when running the currently running code fragment based on the current running data; and displaying the monitoring variable value in a fifth display area. Therefore, it is easier to understand the current running status of the model by displaying the values of the monitoring variables during the current runtime.
  • the method 100 further includes: determining a model output result, which is generated by running the currently running code fragment based on the current running data; generating a graphical file of the model output result; in the sixth display area Display graphic files. It can be seen that by graphically displaying the model output results, it is easier to intuitively understand the model effect.
  • Figure 2 is an exemplary schematic diagram illustrating the display effect of an AI model according to an embodiment of the present invention.
  • model basic information 211 is displayed in the first display area 21 .
  • the model basic information 211 may include the registration time of the AI model, model source, distributed version control system comment (GIT comment), user, saving time and model status, etc.
  • the first display area 21 also displays graphical files related to the AI model (for example, graphical files obtained from a graph database).
  • the graphical files contain a hierarchical system between predetermined parameters in the AI model, where the hierarchical system is Obtained based on the model configuration file of the parsed AI model.
  • the first display area 21 also displays key parameter nodes 50 and the specific key parameters pointed to by the key parameter nodes 50 .
  • specific key parameters include: number of iterations 51, version record 52, blocking value 53 and random status 54.
  • the number of iterations 51 points to the default value field 55 (for example, containing the string "default iterations is 1000 times") and the description field 56 for the number of iterations (for example, containing the string "impact: accuracy and computing performance”).
  • the interception value (intercept) 53 points to the default value field 57 (for example, containing the string "Default value: 0.5; range (0,1)”) and the introduction description field 56 for the interception value (for example, containing the string "compensation The impact on blocking accuracy is uncertain”).
  • description information of multiple data sets of the AI model is displayed.
  • the description information includes the name of the dataset displayed in text form, the size of the dataset, the number of labels, etc.
  • the description can also include data from the dataset displayed graphically, usually in the form of a time series. For example, by triggering the "chart" hyperlink in the second display area 22, the data in the corresponding data set can be displayed in a graphical form.
  • the user can select a specific data set in the second display area 22 and use the selected database to train or test the AI model.
  • Parsing the model configuration file of the AI model can obtain the dependent libraries of the AI model. Create a running environment for the AI model based on dependent libraries. Deserialize the AI model to obtain the model code.
  • the model code contains code snippets for processing input data and comments on the code snippets.
  • the AI model can then be run in the runtime environment based on the selected data set. During the running process, it usually manifests itself as the sequential execution of code fragments contained in the AI model.
  • the currently running code fragment 231 of the AI model and the comments 232 of the code fragment 231 are displayed.
  • the current running data provided from the selected data set to the currently running code fragment is synchronously displayed in the fourth display area 24 in a graphical manner, for example, the current running data is displayed in the current sliding window 241 .
  • monitoring variable values when the currently running code fragment displayed in the third display area 23 is executed based on the current running data in the current sliding window 241 is displayed.
  • monitoring variable values can be window size and feature values.
  • the characteristic value may include a characteristic peak value, a characteristic average value, or a characteristic effective value, etc.
  • Displayed in the sixth display area 26 a graphical file of the model output result generated by running the currently running code snippet displayed in the third display area 23 using the current running data in the current sliding window 241.
  • the AI model is a classifier model. As time goes by, the data provided to the model is constantly updated, the currently running code fragment 231 in the third display area 23 is updated synchronously, and the classification effect in the sixth display area 26 is also updated synchronously, thereby realizing a linkage Synchronous display of effects.
  • Control 27 can be used to control the starting, pausing, jumping or backtracking of code fragments, and other operations.
  • the control 27 issues a control instruction, the current running data in the current sliding window 241, the current running code fragment 231 in the third display area 23, and the model effect in the sixth display area 26 also change synchronously.
  • the first display area 21, the second display area 22, the third display area 23, the fourth display area 24, the fifth display area 25 and the sixth display area 26 are included in the same display interface, thereby facilitating the user's global Understand the model running status.
  • the embodiment of the present invention optimizes the display effect of the AI model.
  • the model's dataset description information is no longer separate from the model.
  • it realizes visual AI model display, including code snippets and vivid model effect display, and also supports multiple control methods of the model execution process (including training process and testing process).
  • FIG. 3 is an exemplary schematic diagram of a presentation process of an AI model according to an embodiment of the present invention.
  • the display process of the AI model includes a model registration process 31, a storage process 32, and a model display process 33.
  • the user uploads the AI model.
  • the user can further register manager information 311 in the model registration process 31 , add code comments 312 (for example, add comments directly in the model code or add comments through the model configuration file), and set monitoring variables 313 .
  • the user can also upload a data set and register the data set 314, set tags 315, and match the model 316.
  • matching model 316 the data set is associated with the AI model.
  • the stored process 32 includes model storage 321 and data set storage 322.
  • the model storage 321 the AI model uploaded by the user is stored.
  • the data set storage 322 the data set uploaded by the user is stored.
  • the model display process 33 includes model query 331 and data set query 332.
  • the model keyword for example, model name
  • the model keyword is used to query the corresponding AI model from the model storage 321, and the associated AI model is queried from the graph database
  • the graphical file contains a hierarchical system between predetermined parameters in the AI model, where the hierarchical system is obtained based on parsing the model configuration file of the AI model.
  • a dataset query 332 is executed using the model keyword to retrieve a dataset corresponding to the model keyword and description information of the dataset from the dataset storage 332 .
  • the model query 331 sends the retrieved AI model and graphical files to the model presentation process 33 .
  • the data set query 332 sends the retrieved data set and the description information of the data set to the model presentation process 33 .
  • the model display process 33 controls the code management 70 and the display process of the visualization board 29 .
  • the visualization board 29 includes a first display area 21 , a second display area 22 , a third display area 23 , a fourth display area 24 , a fifth display area 25 and a sixth display area 26 .
  • code management 70 deserialization 333 is performed on the model to obtain model code, which includes code snippets and comments for the code snippets.
  • the AI model is run based on the data set retrieved from the data set storage 332 .
  • the data tracking 335 of the model display process 33 is used to output graphical files (including the hierarchy between parameters), currently executed code fragments and their comments, description information of the data set, monitoring variable values, and model output results.
  • the graphical file (including the hierarchy between parameters), the currently executed code fragment and its comments, the description information of the data set, the monitoring variable values and Model output results.
  • the chart display 337 of the time series input data in the fourth display area 24 the current operating data in the data set is graphically displayed.
  • a graphical file of the model output result is generated, and the graphical file of the model output result is displayed in the chart display 338 of the time series output data in the sixth display area 26.
  • the graphical file (including the hierarchical system between parameters) is displayed through the first display area 21 , the description information of the data set is displayed through the second display area 22 , and the current data is displayed through the third display area 23 The executed code fragments and their comments are displayed, as well as the monitored variable values via the fifth display area 25 .
  • Step 1 Query the AI model or data set and graphically display the key parameters of the model, which helps to obtain the correlation between the model parameters and the data set.
  • Step 2 Extract the time series values of the data based on the model's relevant data set.
  • Step 3 Download the dependencies of the AI model and create a model running environment. Therefore, deserialized model scripts can be tested in the embedded IDE.
  • the input (time series data) and output (algorithm analysis process and results) of the AI model are visually displayed, and the linkage display between input and output can be controlled through calculation script operations.
  • Step 4 Data communication and operational control between visualizations and calculation scripts.
  • FIG 4 is an exemplary schematic diagram of the subscription process and query process of the AI model according to the embodiment of the present invention.
  • the subscription process 60 includes: based on parsing the model configuration file of the registered model 40, obtaining the hierarchical system between predetermined parameters in the model 40, converting the hierarchical system into a graphics file 41, and then storing the graphics file 41 to the graph database 42 .
  • the query process includes: receiving the model keyword 43, using the model keyword 43 to retrieve the corresponding model and the data set information 44 associated with the retrieved model. Then, in the graphics file display 45, the graphics file 41 and the data set information 44 are displayed.
  • Figure 5 is an exemplary schematic diagram of tracking model training in a code board and a vision board according to an embodiment of the present invention. As shown in Figure 5:
  • Step 801 Issue a code start instruction via the control 71 that controls operations such as starting, pausing, jumping or backtracking of the code fragment.
  • Step 802 The window provides data 72 to provide current running data.
  • Step 803 The model calculation process 73 uses the current running data to execute the current running code and obtain the current model output result.
  • Step 804 The image rendering process 75 caches the model output result.
  • Step 805 The image rendering process 75 sends the cached model output result to the rendering layer to display the graphical model output result.
  • FIG. 6 is an exemplary structural diagram of a display device of an AI model according to an embodiment of the present invention.
  • the device 600 includes: a receiving module 601, configured to receive a model query request containing a model keyword; a determining module 602, configured to determine an artificial intelligence model corresponding to the model keyword based on the model query request;
  • the acquisition module 603 is configured to obtain a graphical file associated with the artificial intelligence model, where the graphical file contains a hierarchical system between predetermined parameters in the artificial intelligence model, and the hierarchical system is obtained based on analyzing the model configuration file of the artificial intelligence model. ;
  • Display module 604 configured to display graphical files in the first display area.
  • the determination module 602 is configured to determine the data set of the artificial intelligence model, the data set includes a training data set or a test database; determine the text description information and graphical description information of the data set, and the graphical description information is adapted to Display the data in the data set graphically; the display module 604 is configured to display text description information and/or graphical description information in a second display area surrounding the first display area.
  • the acquisition module 603 is configured to obtain the dependency library of the artificial intelligence model; create a running environment for the artificial intelligence model based on the dependency library; and deserialize the artificial intelligence model to obtain the model code.
  • the model code Contains code snippets and comments for the code snippets; in the running environment, runs the artificial intelligence model based on the selected data set; the display module 604 is configured to display the currently running code snippets and their comments in the third display area, in In the fourth display area, the current running data provided from the selected data set to the currently running code fragment is synchronously displayed in a graphical manner.
  • the acquisition module 603 is configured to determine the monitoring variable value when running the currently running code fragment based on the current running data; the display module 604 is configured to display the monitoring variable value in the fifth display area.
  • the acquisition module 603 is configured to determine the model output result, which is generated by running the currently running code fragment based on the current running data; generate a graphical file of the model output result; the display module 604, is configured to display the graphical file in the sixth display area.
  • the implementation of the present invention is helpful for users to understand the AI models and data sets hosted in MLOps, facilitate intuitive observation of the AI models, and directly observe the running process of the models using data analysis methods.
  • the embodiment of the present invention also supports a linked playback mode to save computing resources. Data transfer and model output can be linked via debug mode embedded in the page.
  • the play window mode can realize segmented observation and playback mode of data, which is beneficial to tracking, understanding and analysis.
  • the present invention provides an intuitive and vivid model display method, which is helpful for AI model management and operation in different use cases.
  • the embodiment of the present invention can also be implemented as a service, such as applied to MindSphere, Industry Edge, Mendix or other software systems, etc.
  • FIG. 7 is an exemplary structural diagram of an electronic device according to an embodiment of the present invention.
  • the electronic device 700 includes a processor 701, a memory 702, and a computer program stored on the memory 702 and executable on the processor 701.
  • the computer program is executed by the processor 701
  • any of the above AI models are implemented. display method.
  • the memory 702 can be implemented as various storage media such as electrically erasable programmable read-only memory (EEPROM), flash memory (Flash memory), programmable programmable read-only memory (PROM), etc.
  • Processor 701 may be implemented to include one or more central processing units or one or more field programmable gate arrays, where a field programmable gate array integrates one or more central processing unit cores.
  • the central processing unit or central processing unit core may be implemented as a CPU, an MCU, a DSP, or the like.
  • each step is not fixed and can be adjusted as needed.
  • the division of each module is only for the convenience of describing the functional division. In actual implementation, one module can be implemented by multiple modules, and the functions of multiple modules can also be implemented by the same module. These modules can be located on the same device. , or it can be on a different device.
  • the hardware modules in various embodiments may be implemented mechanically or electronically.
  • a hardware module may include specially designed permanent circuits or logic devices (such as a dedicated processor such as an FPGA or ASIC) to perform specific operations.
  • Hardware modules may also include programmable logic devices or circuits (eg, including general-purpose processors or other programmable processors) temporarily configured by software to perform specific operations.
  • programmable logic devices or circuits eg, including general-purpose processors or other programmable processors

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

Des modes de réalisation de la présente invention concernent un procédé et un appareil d'affichage de modèle d'intelligence artificielle (IA), un dispositif électronique et un support de stockage. Le procédé consiste à : recevoir une demande d'interrogation de modèle contenant un mot-clé de modèle ; sur la base de la demande d'interrogation de modèle, déterminer un modèle d'IA correspondant au mot-clé de modèle ; obtenir un fichier graphique associé au modèle d'IA, le fichier graphique contenant un système hiérarchique de paramètres prédéfinis dans le modèle d'IA, et le système hiérarchique étant obtenu par analyse d'un fichier de configuration de modèle du modèle d'IA ; et afficher le fichier graphique dans une première zone d'affichage. Dans les modes de réalisation de la présente invention, le système hiérarchique des paramètres dans le modèle d'IA est affiché dans un mode graphique, de telle sorte que l'efficacité d'affichage du modèle est améliorée, et un utilisateur peut connaître profondément les performances de modèle. De plus, les fragments de code en cours d'exécution et les données de fonctionnement en cours sont affichés de manière synchrone dans un mode graphique, ce qui facilite la compréhension des significations des paramètres et de l'effet de fonctionnement de modèle.
PCT/CN2022/116490 2022-09-01 2022-09-01 Procédé et appareil d'affichage de modèle d'intelligence artificielle, dispositif électronique et support de stockage WO2024045128A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/116490 WO2024045128A1 (fr) 2022-09-01 2022-09-01 Procédé et appareil d'affichage de modèle d'intelligence artificielle, dispositif électronique et support de stockage

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/116490 WO2024045128A1 (fr) 2022-09-01 2022-09-01 Procédé et appareil d'affichage de modèle d'intelligence artificielle, dispositif électronique et support de stockage

Publications (1)

Publication Number Publication Date
WO2024045128A1 true WO2024045128A1 (fr) 2024-03-07

Family

ID=90100150

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/116490 WO2024045128A1 (fr) 2022-09-01 2022-09-01 Procédé et appareil d'affichage de modèle d'intelligence artificielle, dispositif électronique et support de stockage

Country Status (1)

Country Link
WO (1) WO2024045128A1 (fr)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110689124A (zh) * 2019-09-30 2020-01-14 北京九章云极科技有限公司 一种构建神经网络模型的方法及系统
CN110782031A (zh) * 2019-09-27 2020-02-11 北京计算机技术及应用研究所 多框架卷积神经网络模型结构可视化以及网络重建方法
CN112883242A (zh) * 2021-04-28 2021-06-01 成都新希望金融信息有限公司 树形机器学习模型可视化方法、装置、电子设备及存储介质

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110782031A (zh) * 2019-09-27 2020-02-11 北京计算机技术及应用研究所 多框架卷积神经网络模型结构可视化以及网络重建方法
CN110689124A (zh) * 2019-09-30 2020-01-14 北京九章云极科技有限公司 一种构建神经网络模型的方法及系统
CN112883242A (zh) * 2021-04-28 2021-06-01 成都新希望金融信息有限公司 树形机器学习模型可视化方法、装置、电子设备及存储介质

Similar Documents

Publication Publication Date Title
EP3508981B1 (fr) Plate-forme de test sans contact
EP3433732B1 (fr) Conversion de diagrammes visuels en code
US20200125482A1 (en) Code examples sandbox
US10210211B2 (en) Code searching and ranking
WO2022041406A1 (fr) Procédé de surveillance de violations d'applications basé sur une reconnaissance optique de caractères et sur un apprentissage par transfert
US9852217B2 (en) Searching and ranking of code in videos
US10489124B2 (en) Method and system for providing software containers supporting executable code created from computational algorithms described in printed publications
US10509719B2 (en) Automatic regression identification
CN115827895A (zh) 一种漏洞知识图谱处理方法、装置、设备及介质
CN107391528B (zh) 前端组件依赖信息搜索方法及设备
Bajammal et al. A survey on the use of computer vision to improve software engineering tasks
JP2017539031A (ja) テスト実行からのテスト検証の分離
US10365995B2 (en) Composing future application tests including test action data
CN114416481A (zh) 日志分析方法、装置、设备及存储介质
WO2017164856A1 (fr) Identifications d'objets d'interface utilisateur comparables
CN113886204A (zh) 用户行为数据收集方法、装置、电子设备及可读存储介质
CN106484389B (zh) 动作流分段管理
WO2016190869A1 (fr) Détermination d'actions de test potentielles
Becker et al. Core technologies for the internet of services
WO2024045128A1 (fr) Procédé et appareil d'affichage de modèle d'intelligence artificielle, dispositif électronique et support de stockage
JP2017522639A5 (fr)
US11544179B2 (en) Source traceability-based impact analysis
US11315436B2 (en) Method and system for automatically providing graphical user interfaces for computational algorithms described in printed publications
CN115599520A (zh) 一种数据智能处理方法、装置、设备及存储介质
CN115481025A (zh) 自动化测试的脚本录制方法、装置、计算机设备及介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22956953

Country of ref document: EP

Kind code of ref document: A1