Artificial intelligence training system and method

Abstract

The application provides an artificial intelligence practical training system and method, comprising a cloud management platform, a cloud management module and a teaching module; wherein the cloud management platform pre-stores a plurality of experiment environment images, the experiment environment images comprising container environment images and virtual machine environment images; the teaching module is used for responding to experiment requests of teacher users or student users; the cloud management module is used for sending experiment environment scheduling requests to the cloud management platform based on the experiment requests; the cloud management platform is used for receiving the experiment environment scheduling requests sent by the cloud management module, determining target experiment environment images based on the experiment environment scheduling requests, and sending the target experiment environment images to the teaching module, so that the teacher users or student users can perform experiment training. The application can flexibly provide container environment images or virtual machine environment images for users, and improves the flexibility and universality of the artificial intelligence practical training system.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence training system and method. Background Technology

[0002] Existing big data or artificial intelligence experimental training systems typically provide experimental environments based on containers or virtual machines. Their overall architecture generally includes one or more management nodes and one or more computing nodes. These systems provide users with containers or virtual machines for teaching or research, and users can access the experimental environment through a browser or a dedicated client. Currently, AI-related experiments include both Linux-based and Windows-based environments, but existing experimental training systems usually only support one of these methods: containers or virtual machines. Therefore, existing AI experimental training systems have limited functionality and poor versatility. Summary of the Invention

[0003] In view of this, the purpose of this invention is to provide an artificial intelligence training system and method that can flexibly provide users with container environment images or virtual machine environment images, thereby improving the flexibility and versatility of the artificial intelligence training system.

[0004] To achieve the above objectives, the technical solutions adopted in the embodiments of the present invention are as follows:

[0005] In a first aspect, embodiments of the present invention provide an artificial intelligence training system, comprising: a cloud management platform, a cloud management module, and a teaching module; wherein, the cloud management platform pre-stores multiple experimental environment images, including: container environment images and virtual machine environment images; the teaching module is used to respond to experimental requests from teacher users or student users; the cloud management module is used to send experimental environment scheduling requests to the cloud management platform based on the experimental requests; the cloud management platform is used to receive the experimental environment scheduling requests sent by the cloud management module, determine the target experimental environment image based on the experimental environment scheduling requests, and send the target experimental environment image to the teaching module so that teacher users or student users can conduct experimental training.

[0006] In one implementation, the cloud management platform includes: a container management node, a virtual machine management node, and at least one compute node; wherein the container management node is used to store and manage container environment images, the virtual machine management node is used to store and manage virtual machine environment images, and the compute node is used to provide containers or virtual machines for container environment images or virtual machine environment images.

[0007] In one implementation, the teaching module includes: a student center, a teacher center, and an examination center; wherein the student center includes at least: an online demonstration viewing unit, an attendance check-in unit, an online Q&A answering unit, an experiment unit, an experiment report writing unit, an experiment recording unit, a file upload unit, a file download unit, and a file browsing unit; the teacher center includes at least: a lesson preparation unit, a course management unit, an experiment demonstration unit, and a monitoring unit; and the examination center includes at least: a test question management unit, an exam customization unit, an online examination unit, and an intelligent analysis unit.

[0008] In one embodiment, the system further includes an online teaching module; wherein the online teaching module includes at least: a screen sharing unit, an online screen recording unit, an online attendance unit, a random questioning unit, an experiment monitoring unit, and a teaching tool unit.

[0009] In one embodiment, the system further includes: an artificial intelligence module for demonstrating artificial intelligence models; wherein the artificial intelligence models include at least: a gender and age recognition model, an image style transfer model, a human keypoint recognition model, a facial expression recognition model, an image description model, and a YOLO object detection model.

[0010] In one embodiment, the system further includes a timed task module for scheduling demonstration tasks of artificial intelligence models in the artificial intelligence module.

[0011] In one implementation, the system further includes a course module for creating new experimental environment images and managing courses.

[0012] In one implementation, the system further includes a messaging module for teacher or student users to send and receive messages.

[0013] In one embodiment, the system further includes a main control module and a common module; wherein the main control module is connected to the cloud management module, the teaching module, the online teaching module, the artificial intelligence module, the scheduled task module, the course module, the message module, and the common module.

[0014] Secondly, embodiments of the present invention provide an artificial intelligence training method, which is applied to any of the artificial intelligence training systems provided in the first aspect. The method includes: responding to an experiment request from a teacher or student; sending an experiment environment scheduling request to a cloud management platform based on the experiment request; determining a target experiment environment image based on the experiment environment scheduling request; and sending the target experiment environment image to a teaching module so that the teacher or student can conduct experiment training.

[0015] The embodiments of the present invention bring the following beneficial effects:

[0016] The artificial intelligence training system and method provided in this embodiment of the invention include: a cloud management platform, a cloud management module, and a teaching module. The cloud management platform pre-stores multiple experimental environment images, including container environment images and virtual machine environment images. The teaching module responds to experimental requests from teachers or students. The cloud management module sends experimental environment scheduling requests to the cloud management platform based on the experimental requests. The cloud management platform receives the experimental environment scheduling requests from the cloud management module, determines the target experimental environment image based on the requests, and sends the target experimental environment image to the teaching module so that teachers or students can conduct experimental training. The system pre-stores multiple experimental environment images (container environment images and virtual machine environment images) on the cloud management platform, supporting both container-based Linux experimental environments and virtual machine-based Windows or Linux experimental environments. It can schedule container environment images or virtual machine environment images from the cloud management platform according to the experimental needs of teachers or students, thereby flexibly providing users with container environment images or virtual machine environment images and improving the flexibility and versatility of the artificial intelligence training system.

[0017] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0019] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the structure of an artificial intelligence training system provided in an embodiment of the present invention;

[0021] Figure 2 A schematic diagram of a cloud management platform provided in an embodiment of the present invention;

[0022] Figure 3 This is a schematic diagram of another artificial intelligence training system provided in an embodiment of the present invention;

[0023] Figure 4A flowchart for creating an experimental environment is provided as an embodiment of the present invention;

[0024] Figure 5 A module flowchart of an artificial intelligence training system provided in an embodiment of the present invention;

[0025] Figure 6 A flowchart illustrating the use of an artificial intelligence training system provided in an embodiment of the present invention;

[0026] Figure 7 A flowchart of an artificial intelligence training method provided in an embodiment of the present invention;

[0027] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0028] icon:

[0029] 10-Cloud Management Platform; 20-Cloud Management Module; 30-Teaching Module; 40-Online Teaching Module; 50-Artificial Intelligence Module; 60-Scheduled Task Module; 70-Course Module; 80-Message Module; 90-Main Control Module; 100-Common Module. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] Currently, existing artificial intelligence training systems provide great convenience for experimental training in big data and artificial intelligence, but they also have the following drawbacks: (1) Although most big data and artificial intelligence-related experiments are based on the Linux environment, some experiments require the Windows environment. However, experimental training systems based on container technology can only support the Linux environment due to technical reasons and cannot support the Windows environment. (2) Experimental training systems based on virtual machine technology can use virtual machines based on both Linux and Windows systems, but the startup speed of virtual machines is much slower than that of containers. Containers can start in seconds, while virtual machines take several minutes. When multiple people start virtual machines at the same time, the startup time is even longer, which seriously affects the teaching experience. (3) Common experimental training systems can only choose to use either container technology or virtual machine technology and cannot use them simultaneously. (4) Common experimental training systems only provide experimental functions and do not have teaching functions. Teachers need to use other systems to show courseware and demonstrate experimental processes to students. (5) Chapter experiments are not independent. To do the later experiments, all the previous experiments must be completed. Therefore, existing artificial intelligence experimental training systems have limited functions and poor versatility.

[0032] Based on this, the artificial intelligence training system and method provided in this embodiment of the invention can flexibly provide users with container environment images or virtual machine environment images, thereby improving the flexibility and versatility of the artificial intelligence training system.

[0033] To facilitate understanding of this embodiment, a detailed description of an artificial intelligence training system disclosed in this embodiment of the invention will be provided first. (See [link to relevant documentation]). Figure 1 The diagram shows the structure of an artificial intelligence training system, illustrating that the system mainly includes the following components: a cloud management platform 10, a cloud management module 20, and a teaching module 30. The cloud management platform 10 pre-stores multiple experimental environment images, including container environment images and virtual machine environment images. The teaching module 30 responds to experimental requests from teachers or students. The cloud management module 20 sends experimental environment scheduling requests to the cloud management platform based on the experimental requests. The cloud management platform 10 receives the experimental environment scheduling requests from the cloud management module 20, determines the target experimental environment image based on the scheduling requests, and sends the target experimental environment image to the teaching module 30 so that teachers or students can conduct experimental training.

[0034] Specifically, the artificial intelligence training system provided in this embodiment of the invention integrates Docker engine technology, Kubernetes container orchestration engine technology, and OpenStack cloud platform technology. It can flexibly provide users with containers or virtual machines according to teaching and research needs. The system can provide more than 200 container and virtual machine experimental images and offers independent experimental environments for all experiments. Each experiment can be conducted independently without requiring all previous experiments to be completed before starting the next experiment. Users do not need to install a dedicated client; they only need to log in to the system using a browser, select the desired experiment through the teaching module 30, and the cloud management module 20 communicates with the cloud management platform 10 to send the experiment request. The cloud management platform 10 then schedules the target experimental environment image required for the experiment based on the request, allowing users to access the desired experimental environment.

[0035] The artificial intelligence training system provided in this embodiment of the invention has multiple experimental environment images (container environment images and virtual machine environment images) pre-stored on the cloud management platform. It supports both container-based Linux experimental environments and virtual machine-based Windows or Linux experimental environments. It can schedule container environment images or virtual machine environment images from the cloud management platform according to the experimental needs of teachers or students, thereby flexibly providing users with container environment images or virtual machine environment images and improving the flexibility and versatility of the artificial intelligence training system.

[0036] In one implementation, see Figure 2 As shown, the cloud management platform includes: a container management node, a virtual machine management node, and at least one compute node; wherein, the container management node is used to store and manage container environment images, the virtual machine management node is used to store and manage virtual machine environment images, and the compute node is used to provide containers or virtual machines for container environment images or virtual machine environment images.

[0037] Specifically, the container management node can provide container orchestration services based on Docker engine technology and Kubernetes container orchestration engine technology. It is responsible for providing web, database, remote dictionary service (Redis), nginx load balancing service and network file system (NFS) sharing service for the artificial intelligence training system, including container image management, container management, cluster management, host management and system monitoring functions.

[0038] The virtual machine management node can provide virtual machine orchestration services based on Docker engine technology and the OpenStack cloud platform, including virtual machine management, virtual machine image management, network management, security group management, floating IP management, instance type management, project management, user management, quota management, system monitoring and other functions.

[0039] The compute nodes are managed simultaneously by container management nodes and virtual machine management nodes. They can provide either containers or virtual machines based on Docker engine technology, Kubernetes container orchestration engine technology, the OpenStack cloud platform, and keyboard / video / mouse (KVM) ports, depending on the experimental needs. In practice, once the experimental environment is determined based on specific experimental requirements, the container management node or virtual machine management node can schedule the compute nodes to provide containers or virtual machines to complete the experiment based on their load status.

[0040] In one implementation, the aforementioned teaching module includes: a student center, a teacher center, and an examination center; wherein the student center includes at least: an online demonstration viewing unit for watching online experiment demonstrations, an online attendance unit for online attendance, an online question answering unit for answering questions, an experiment unit for conducting experiments, an experiment report writing unit for writing experiment reports, an experiment recording unit for recording experiment data, a file upload unit for uploading, downloading, and browsing files using cloud storage functionality, a file download unit, and a file browsing unit; the teacher center includes at least: a lesson preparation unit for lesson preparation, a course management unit for managing course data, materials, report templates, and college-related content, an experiment demonstration unit for demonstrating experiments, and a monitoring unit for monitoring students' experiment processes; the examination center includes at least: a test question management unit for managing test questions, an examination customization unit for customizing examination content, an online examination unit for examinations, and an intelligent analysis unit for examination analysis.

[0041] Specifically, student and teacher users can access the AI ​​training system through the web service of the container management node. Students can use the student center to learn courses required by their teachers, watch online demonstrations, check in online, answer online questions, enter experimental environments to conduct experiments according to requirements, write experimental reports online, record experimental notes online, upload, download, and browse their own files using the cloud storage function, participate in online exams using the exam center, view their learning path, exam analysis, check-in analysis, and experimental report analysis using the intelligent analysis unit, and experience deep learning and machine learning related models using the model lobby.

[0042] Teachers can customize courses, manage datasets, materials, and report templates in the lesson preparation unit, manage courses and colleges in the teaching center, teach online, demonstrate experimental content, monitor students' experimental processes, manage test questions and customize exams in the examination center, use the intelligent analysis unit to view students' learning paths, exam analysis, attendance analysis, experimental report analysis, and use the model lobby to experience deep learning and machine learning related models.

[0043] Optionally, the system also includes a resource center, where administrators can monitor the hardware resources used by compute node containers and virtual machines after accessing the system; manage files, image tags, images, virtual machines, containers, snapshots, disks, tasks, recordings, and report templates globally in the resource center; manage user, class, and organization information in system management; view online users, login logs, and operation logs in system monitoring; and manage database backups, set system personalization information, manage authorizations, and manage notifications in operation and maintenance management.

[0044] See Figure 3 The diagram shown illustrates the structure of another artificial intelligence training system, demonstrating how the system... Figure 1 In addition to this, it also includes: online teaching module 40, artificial intelligence module 50, scheduled task module 60, course module 70, message module 80, main control module 90, and public module 100.

[0045] In one embodiment, the online teaching module 40 includes at least: a screen sharing unit, an online screen recording unit, an online attendance unit, a random questioning unit, an experiment monitoring unit, and a teaching tool unit. Specifically, the system provided in this embodiment supports online teaching functions, enabling practical online teaching functions such as screen sharing, online screen recording, online attendance, and online delivery. The system includes buttons for these functions, allowing users to perform the corresponding functions by clicking the buttons. In specific implementations, the screen sharing unit allows students to view the teacher's shared canvas, PPT, or video through a browser; the online screen recording unit records the teaching process, allowing teachers to record online in a browser and share the video for students to download after class; the online attendance unit allows online students to complete attendance within a specified time, and teachers can view student attendance data; the random questioning unit randomly selects a student to answer a question; the experiment monitoring unit monitors students' experiment screens; and the teaching tool unit provides teaching environments such as containers, virtual machines, Jpyter, deep learning, and machine learning.

[0046] In one implementation, the artificial intelligence module 50 is used to demonstrate artificial intelligence models; wherein, the artificial intelligence models include at least: a gender and age recognition model, an image style transfer model, a human keypoint recognition model, a facial expression recognition model, an image description model, and a YOLO object detection model. Specifically, the system provides a model lobby function, that is, it provides demonstrations of artificial intelligence models to users through the artificial intelligence module. In this embodiment of the invention, the artificial intelligence models are pre-trained, and users can input images into the artificial intelligence models during the experience to obtain corresponding results.

[0047] (1) The gender and age recognition model was trained using TensorFlow (a symbolic mathematics system based on dataflow programming) and the imdb_crop and wiki_crop face datasets. The experiment first used dlib (a toolkit written in C++) to perform face detection and face alignment on the images, and then used deep convolutional neural networks (CNNs) to estimate the age and gender of people, and used the FaceNet network structure to extract features. Pre-trained models in the FaceNet network were used to accelerate network training.

[0048] Gender prediction is a binary classification problem. The output layer of the gender prediction network has two classes, with two nodes representing "male" and "female". The maximum value of these two outputs is taken as the final gender. Ideally, age prediction can be treated as a regression problem, but accurately estimating age through regression is challenging. Therefore, age prediction can be described as a classification problem, estimating the age group. For example, ages in the range of 0-2 are one class, 4-6 are another class, and so on, dividing the dataset into the following 8 age groups [(0-2), (4-6), (8-12), (15-20), (25-32), (38-43), (48-53), (60-100)].

[0049] (2) The image style transfer model was trained using TensorFlow and the VGG16 model. A total of 7 models were trained using the COCO2014 dataset. The image style transfer model uses two networks: a loss network and a generation network. The loss network is the VGG16 model, which can be implemented using TensorFlow. The image generation network can be defined according to requirements. The principle of the image generation network is to first perform convolution calculation on the image, and then perform "deconvolution" calculation, which is equivalent to encoding the image and then restoring it to an image, outputting the synthesized image data.

[0050] (3) The implementation principle of the human keypoint recognition model is OpenPose human pose recognition. The OpenPose human pose recognition project is an open-source library developed based on convolutional neural networks and supervised learning, using the Convolutional Architecture for Fast Feature Embedding (Caffe) framework. It can estimate poses such as human movements, facial expressions, and finger movements. In this embodiment of the invention, the COCO dataset can be used for model training. The COCO dataset is designed specifically for object detection, segmentation, human keypoint detection, semantic segmentation, and caption generation, containing approximately 25GB of images and approximately 600MB of label files. The data is preprocessed into heatmaps and vector maps, and the keypoints of the human images are labeled. The processed images are then used for model training. In this embodiment of the invention, OpenPose can be implemented using TensorFlow. The model is trained 59,000 times on a server to obtain the TensorFlow model.

[0051] (4) The face expression recognition model mainly uses the dlib library and the pre-trained face feature point 68 model, as shown in the following model: shape_predictor_68_face_landmarks.dat. OpenCV is used to operate on the image to recognize the face and mark the feature points.

[0052] The recognition rules include: 1) The larger the proportion of the mouth opening distance to the width of the facial recognition frame, the more excited the emotion, which may be very happy or extremely angry. 2) Raised eyebrows: The smaller the ratio of the distance from feature points 17-21 or 22-26 from the top of the facial recognition frame to the height of the frame, the more pronounced the eyebrow raising, which can indicate surprise or happiness; the angle of eyebrow tilt: when happy, eyebrows are generally raised, while when angry, eyebrows are furrowed and lowered more significantly. 3) Squinting eyes: people unconsciously squint when laughing heartily, and widen their eyes when angry or surprised.

[0053] (5) The image captioning model is Google's deep learning model im2txt, trained using the 2015 MSCOCO ImageCaptioning Challenge dataset. This model is an encoder-decoder model. The encoder is a CNN model commonly used in image recognition, object detection, and other fields. Various common convolutional networks can be used, such as VGG, Inception, ResNet, etc. In this embodiment of the invention, the Inception-V3 model trained on the ILSVRC-2012-CLS dataset can be used. Inputting the image into Inception-V3 will yield a fixed-length vector, which can be seen as the feature extracted from the image. The decoder is an LSTM model commonly used in language models or machine translation. The fixed-length vector output from the encoder is input into the decoder to obtain a description of the image.

[0054] (6) YOLO Object Detection Model: YOLO, short for You Only Look Once, is an end-to-end object detection method, surpassing many previous algorithms. As a deep neural network, YOLO does not require preprocessing of images, extraction of relevant features, object localization, or image segmentation. In the inference stage, it can directly input color images and output the target or object classification and its coordinates, hence the term end-to-end. The YOLOv3 base network uses Darknet-53, and it also uses a single network evaluation for prediction, making it extremely fast—more than 1000 times faster than R-CNN and 100 times faster than Fast R-CNN.

[0055] The training dataset used is VOC2007, which contains 9963 labeled images and consists of three parts: train, val, and test. A total of 24,640 objects are labeled. The downloaded dataset annotations are in XML format, which can be converted into txt format before training.

[0056] In one implementation, the timed task module 60 can be used to schedule demonstration tasks of the artificial intelligence model in the artificial intelligence module 50.

[0057] In one implementation, the course module 70 can be used to create new experimental environment images and manage courses. Specifically, the system provides a course workshop function, allowing users to customize new experimental environments based on any existing experimental environment image through the course module 70. See also Figure 4 The flowchart shown here illustrates a process for creating an experimental environment, which mainly includes the following steps (1) to (10):

[0058] Step (1), create the image.

[0059] Specifically, you can choose to create either a container image or a virtual machine image based on your experimental needs.

[0060] Step (2): If you choose to process a container image, call the openmakeimageinstance interface to start the container environment in the k8s cluster.

[0061] Step (3): Save the container image by calling the savemakeimageinstance interface to save the running container as a new Docker image.

[0062] Step (4): Upload the container image repository by calling the DockerClientUtils.pushImageToNode interface to upload the new Docker image that has been saved to the Harbor image repository.

[0063] Step (5) distribute to compute nodes by calling the DockerClientUtils.pullImageToAllNodes interface to pull the newly created image to each compute node.

[0064] Step (6): If you choose to process a virtual machine image, call the openmakeimageinstance virtual machine interface to start the virtual machine on the OpenStack cluster.

[0065] Step (7): Create a virtual machine snapshot by calling the savemakeimageinstance virtual machine interface to create a snapshot based on the virtual machine.

[0066] Step (8): Upload the virtual machine image service, call the openstackglance interface, and save the virtual machine snapshot to the openstack image service storage.

[0067] Step (9): Release excess resources by calling the OpenStack4jUtils.deleteServer interface to delete virtual machines that have already been snapshotted and release computing resources.

[0068] Step (10): Synchronize the image data, call the MariaDB database interface, and add the newly created image information.

[0069] In one implementation, the messaging module 80 can be used by teacher users or student users to send and receive messages.

[0070] In one implementation, the main control module 90 is connected to the cloud management module 20, the teaching module 30, the online teaching module 40, the artificial intelligence module 50, the scheduled task module 60, the course module 70, the message module 80, and the public module 100. The main control module 90 manages all of the above modules, and the public module 100 provides utility classes for all modules.

[0071] See Figure 5 The flowchart shown here illustrates the modules of an artificial intelligence training system, which include: a cloud management module (tms-module-cloud) 20, a teaching module (tms-module-teaching) 30, an online teaching module (tms-module-online) 40, an artificial intelligence module (tms-module-aiplatform) 50, a scheduled task module (tms-module-quartz) 60, a course module (tms-module-course) 70, a message module (tms-module-message) 80, a main control module (tms-module-admin) 90, and a common module (tms-module-common) 100.

[0072] The system is divided into several modules: a public module providing utility classes for all modules; a messaging module responsible for inter-business message communication; a scheduled task module providing task scheduling services for the artificial intelligence module; an artificial intelligence module providing neural network and data preprocessing algorithms, as well as training and testing of machine learning algorithms for deep learning and machine learning; an online teaching module providing online teaching functions, including screen sharing, online screen recording, online attendance, and online delivery; a course module providing course management functions; a cloud management module providing cloud service scheduling for students and teachers to open experimental environments; a teaching module supporting the entire teaching service and providing overall functions for students and teachers; and a main control module managing all the above modules.

[0073] For ease of understanding, this embodiment of the invention also provides a usage flow for an artificial intelligence training system, see [link to documentation]. Figure 6 As shown, the steps include S601 to S608 as follows:

[0074] Step S601: Teacher users create courses.

[0075] Specifically, after logging into the system, teachers can create courses, upload lab manuals, courseware, and customize course structures according to their needs.

[0076] Step S602: Define the student experimental environment.

[0077] Specifically, teachers can define the experimental environment required for each experiment based on their experimental needs.

[0078] Step S603: Create a container environment or a virtual machine environment.

[0079] Specifically, the system provided in this embodiment of the invention can simultaneously support container environments and virtual machine environments, and can create experimental environments according to the number of environments, CPU and memory quotas, and whether a GPU is required.

[0080] Step S604: Students start the experimental environment.

[0081] Specifically, after logging into the system and entering the relevant course, students can open the experiment they want to do and start the experimental environment.

[0082] Step S605: Perform the experiment according to the experimental manual.

[0083] Specifically, once in the experimental environment, students can complete the experimental steps according to the contents of the experimental manual.

[0084] Step S606: Submit the test report.

[0085] Specifically, while conducting the experimental procedures, students can write and submit experimental reports online. The experimental reports support online screenshots or import Word documents.

[0086] Step S607: Determine whether to save the experimental environment.

[0087] Specifically, choose whether to save the experimental environment. If you choose to save the experimental environment, return to step S604, and then repeat steps S604 to S606 for the rest of the course.

[0088] Step S608: Delete the experimental environment and complete the experiment.

[0089] Specifically, if the experimental environment is not to be saved, it should be deleted to free up hardware resources.

[0090] The artificial intelligence training system provided in this embodiment of the invention has the following beneficial effects:

[0091] (1) By using the flannel network plugin technology, integrating the Kubernetes container orchestration engine technology and the OpenStack cloud platform technology, the problem of network conflict within the same host was solved, enabling the same host to support both container-based Linux experimental environments and virtual machine-based Windows or Linux experimental environments, and both container-based and virtual machine-based experimental environments support GPU calls.

[0092] (2) By recompiling the Linux kernel independently, the original SSD caching function is added. By using small SSD hard disks, a high-speed caching function is provided for large-capacity slow mechanical hard disks, which greatly improves the startup speed of the virtual machine experimental environment. The startup time can be shortened to more than one-tenth of the original time. At the same time, hardware costs are taken into account to avoid using high-priced SSD hard disks.

[0093] (3) Supports online teaching functions, and can realize practical online teaching functions such as screen sharing, online screen recording, online check-in, and online provision.

[0094] (4) Using a B / S architecture design, access is only required via a browser, without the need to install any client software. By modifying the container-based experimental environment to use tigervnc software to implement VNC service, and modifying the OpenStack virtual machine-based experimental environment to use the spice video protocol, the experimental environments of containers and virtual machines, which could only display images at a fixed resolution, have achieved screen resolution adaptive functionality.

[0095] (5) More than 200 container images and virtual machine images are provided. Each image is customized according to the experimental manual of a specific course and chapter, and contains the relevant software and settings required for the corresponding experiment. By setting up experimental environments for different courses and chapters using the corresponding images, chapter experiments are decoupled, allowing students to start learning at any time during the course.

[0096] (6) Added the Course Workshop function, allowing users to customize new experimental environments based on any existing image.

[0097] This invention also provides an artificial intelligence training method, which is applied to the artificial intelligence training system provided in the foregoing embodiments. See [link to relevant documentation]. Figure 7 The flowchart shown illustrates an artificial intelligence training method, which mainly includes the following steps S701 to S703:

[0098] Step S701: Respond to the experiment request from the teacher or student user.

[0099] Step S702: Send an experimental environment scheduling request to the cloud management platform based on the experimental request.

[0100] Step S703: Determine the target experimental environment image based on the experimental environment scheduling request, and send the target experimental environment image to the teaching module so that teachers or students can conduct experimental training.

[0101] The artificial intelligence training method provided in this embodiment of the invention pre-stores multiple experimental environment images (container environment images and virtual machine environment images) on the cloud management platform. It supports both container-based Linux experimental environments and virtual machine-based Windows or Linux experimental environments. It can schedule container environment images or virtual machine environment images from the cloud management platform according to the experimental needs of teachers or students, thereby flexibly providing container environment images or virtual machine environment images to users and improving the flexibility and versatility of the artificial intelligence training system.

[0102] The method provided in this embodiment of the invention has the same implementation principle and technical effect as the aforementioned system embodiment. For the sake of brevity, any part not mentioned in the method embodiment can be referred to the corresponding content in the aforementioned system embodiment.

[0103] This invention also provides an electronic device, specifically, the electronic device includes a processor and a storage device; the storage device stores a computer program, and the computer program, when run by the processor, executes the method described in any of the above embodiments.

[0104] Figure 8 The present invention provides a schematic diagram of the structure of an electronic device 800, which includes a processor 801, a memory 802, a bus 803 and a communication interface 804. The processor 801, the communication interface 804 and the memory 802 are connected through the bus 803. The processor 801 is used to execute executable modules, such as computer programs, stored in the memory 802.

[0105] The memory 802 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 804 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc.

[0106] Bus 803 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0107] The memory 802 is used to store programs. After receiving an execution instruction, the processor 801 executes the program. The method executed by the device for defining the flow process disclosed in any of the foregoing embodiments of the present invention can be applied to the processor 801 or implemented by the processor 801.

[0108] The processor 801 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 801 or by instructions in software form. The processor 801 may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the field, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 802, and processor 801 reads the information in memory 802 and, in conjunction with its hardware, completes the steps of the above method.

[0109] The computer program product of the readable storage medium provided in the embodiments of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the foregoing method embodiments. For specific implementation, please refer to the foregoing method embodiments, which will not be repeated here.

[0110] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0111] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An artificial intelligence training system, characterized in that, include: The cloud management platform comprises a cloud management module and a teaching module; wherein the cloud management platform pre-stores multiple experimental environment images, including container environment images and virtual machine environment images. The teaching module is used to respond to experiment requests from teacher or student users. The cloud management module is used to send an experimental environment scheduling request to the cloud management platform based on the experimental request. The cloud management platform is used to receive the experimental environment scheduling request sent by the cloud management module, determine the target experimental environment image based on the experimental environment scheduling request, and send the target experimental environment image to the teaching module so that the teacher user or student user can conduct experimental training. The cloud management platform includes: a container management node, a virtual machine management node, and at least one compute node; wherein, the container management node is used to store and manage the container environment image, the virtual machine management node is used to store and manage the virtual machine environment image, and the compute node is used to provide containers or virtual machines for the container environment image or the virtual machine environment image; the compute node is managed by the container management node and the virtual machine management node simultaneously, and provides containers or virtual machines based on Docker engine technology, Kubernetes container orchestration engine technology, OpenStack cloud platform, and keyboard, video, or mouse connection ports as needed for the experiment; when the experimental environment is determined according to specific experimental requirements, the container management node or the virtual machine management node schedules the compute node to provide containers or virtual machines according to the load status of the compute node to complete the experiment.

2. The artificial intelligence training system according to claim 1, characterized in that, The teaching modules include: a student center, a teacher center, and an examination center; The student center includes at least the following units: online demonstration viewing unit, attendance check-in unit, online Q&A answering unit, experiment unit, experiment report writing unit, experiment record unit, file upload unit, file download unit, and file browsing unit. The teacher center includes at least: a lesson preparation unit, a course management unit, an experimental demonstration unit, and a monitoring unit; The examination center includes at least: a test question management unit, an examination customization unit, an online examination unit, and an intelligent analysis unit.

3. The artificial intelligence training system according to claim 2, characterized in that, The system further includes an online teaching module; wherein the online teaching module includes at least: a screen sharing unit, an online screen recording unit, an online attendance unit, a random questioning unit, an experiment monitoring unit, and a teaching tool unit.

4. The artificial intelligence training system according to claim 3, characterized in that, The system also includes an artificial intelligence module for demonstrating artificial intelligence models; wherein the artificial intelligence models include at least: a gender and age recognition model, an image style transfer model, a human key point recognition model, a facial expression recognition model, an image description model, and a YOLO object detection model.

5. The artificial intelligence training system according to claim 4, characterized in that, The system also includes a timed task module, used to schedule demonstration tasks of the artificial intelligence model in the artificial intelligence module.

6. The artificial intelligence training system according to claim 5, characterized in that, The system also includes a course module for creating new experimental environment images and managing courses.

7. The artificial intelligence training system according to claim 6, characterized in that, The system also includes a messaging module for the teacher or student to send and receive messages.

8. The artificial intelligence training system according to claim 7, characterized in that, The system further includes a main control module and a common module; wherein the main control module is connected to the cloud management module, the teaching module, the online teaching module, the artificial intelligence module, the scheduled task module, the course module, the message module and the common module.

9. An artificial intelligence training method, characterized in that, The method is applied to the artificial intelligence training system according to any one of claims 1 to 8, and the method includes: Respond to experiment requests from teacher or student users; Based on the experimental request, an experimental environment scheduling request is sent to the cloud management platform; Based on the experimental environment scheduling request, a target experimental environment image is determined and sent to the teaching module so that the teacher or student can conduct experimental training. The cloud management platform includes a container management node, a virtual machine management node, and at least one compute node. The container management node stores and manages the container environment image, the virtual machine management node stores and manages the virtual machine environment image, and the compute node provides containers or virtual machines for the container environment image or the virtual machine environment image. The compute node is managed simultaneously by the container management node and the virtual machine management node, and provides containers or virtual machines based on Docker engine technology, Kubernetes container orchestration engine technology, the OpenStack cloud platform, and keyboard, video, or mouse connection ports, according to experimental needs. Once the experimental environment is determined based on specific experimental requirements, the container management node or the virtual machine management node schedules the compute node to provide containers or virtual machines based on the load status of the compute node to complete the experiment.

Images