WO2020168690A1 - Ai implementation method for classification based on graphical programming tool, and electronic device - Google Patents

Ai implementation method for classification based on graphical programming tool, and electronic device Download PDF

Info

Publication number
WO2020168690A1
WO2020168690A1 PCT/CN2019/100384 CN2019100384W WO2020168690A1 WO 2020168690 A1 WO2020168690 A1 WO 2020168690A1 CN 2019100384 W CN2019100384 W CN 2019100384W WO 2020168690 A1 WO2020168690 A1 WO 2020168690A1
Authority
WO
WIPO (PCT)
Prior art keywords
classification
graphical
predicted
building block
graphical programming
Prior art date
Application number
PCT/CN2019/100384
Other languages
French (fr)
Chinese (zh)
Inventor
李天驰
孙悦
Original Assignee
深圳点猫科技有限公司
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 深圳点猫科技有限公司 filed Critical 深圳点猫科技有限公司
Publication of WO2020168690A1 publication Critical patent/WO2020168690A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F8/00—Arrangements for software engineering
    • G06F8/30—Creation or generation of source code
    • G06F8/34—Graphical or visual programming

Definitions

  • the present invention relates to the field of computer technology, in particular to a classification AI implementation method and electronic equipment based on graphical programming tools.
  • the purpose of the present invention is to provide a classification AI implementation method and electronic equipment based on a graphical programming tool, aiming to solve the problem that the existing programming method does not have an intelligent classification function.
  • a classification AI implementation method based on graphical programming tools including the following steps:
  • the feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
  • a backward propagation algorithm is used to train the graphical classification building blocks.
  • the graphical classification building blocks include an input layer, a hidden layer, and an output layer;
  • the input layer is used to input feature values of things to be predicted
  • the hidden layer is used to extract salient features in the input data, classify them according to the salient features, and transmit the classification results to the output layer;
  • the output layer is used to output classification results.
  • the feature value of the thing to be predicted matches the number of feature values in the training data.
  • the hidden layer is one or more layers, and each hidden layer has a corresponding number of neurons.
  • each set of training data includes feature values and classification results.
  • the uploaded excel data is extracted, the excel is converted into matrix data, and the matrix data is used as training data.
  • An electronic device which includes:
  • a processor suitable for implementing instructions
  • the storage device is adapted to store multiple instructions, and the instructions are adapted to be loaded and executed by the processor:
  • the feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
  • a non-volatile computer-readable storage medium wherein the non-volatile computer-readable storage medium stores computer-executable instructions that, when executed by one or more processors, can cause all The one or more processors execute the classification AI implementation method based on the graphical programming tool.
  • a computer program product wherein the computer program product includes a computer program stored on a non-volatile computer-readable storage medium, and the computer program includes program instructions.
  • the program instructions When the program instructions are executed by a processor, the processor executes all The described classification AI implementation method based on graphical programming tools.
  • the present invention uses graphical classification building blocks to realize the function of a classifier based on the characteristic values of the data type, and packages them into graphical building blocks, which can be directly used in graphical programming tools.
  • graphical building blocks By using these graphical building blocks, users can edit programming works with classified AI, which expands the gameplay and program types of the programmed programs.
  • FIG. 1 is a flowchart of a preferred embodiment of a classification AI implementation method based on a graphical programming tool of the present invention.
  • Figure 2 is a network diagram of an example of a neural network in the present invention.
  • Fig. 3 is a block diagram of an example of a neural network input layer in the present invention.
  • Figure 4 is a block diagram of an example of a hidden layer of a neural network in the present invention.
  • Fig. 5 is a block diagram of an example of a neural network output layer in the present invention.
  • Fig. 6 is a schematic diagram of the training result of a neural network in the present invention.
  • FIG. 7 is a structural block diagram of a preferred embodiment of an electronic device of the present invention.
  • the present invention provides a classification AI implementation method and electronic equipment based on a graphical programming tool.
  • a classification AI implementation method and electronic equipment based on a graphical programming tool.
  • FIG. 1 is a flowchart of a preferred embodiment of a classification AI implementation method based on a graphical programming tool of the present invention. As shown in the figure, it includes the steps:
  • the invention utilizes graphical classification building blocks to realize the function of a classifier based on the characteristic value of the data type, and packs them into graphical building blocks, which can be directly used in graphical programming tools.
  • graphical building blocks By using these graphical building blocks, users can edit programming works with classified AI, which expands the gameplay and program types of the programmed programs.
  • step S1 a classification model is created first.
  • the classification model is similar to a classifier.
  • Classifier is a general term for the method of classifying samples in data mining.
  • the color (wavelength), alcohol concentration and other characteristic values of a bottle of wine can be measured, and the machine judges which one of beer, red wine, and white wine the bottle belongs to based on these characteristic values.
  • Disease judgment The patient goes to the hospital to check the liver function, blood test and other data. These test data are sent to a machine, and the machine judges whether the patient is sick or not based on the data. This kind of machine that can automatically classify the input is called a classifier.
  • the naked eye may be able to make a simple identification of alcohol, but it is unable to judge the disease.
  • a classifier for disease judgment is provided, then only the corresponding detection data needs to be input. Get the judgment result.
  • the classifier based on artificial intelligence can not only complete disease judgments to this degree, it can complete more complicated classification judgments with more feature values.
  • the present invention also mainly adopts the deep learning of neural network to realize the classifier, but obviously for other types of classifiers, it can also be applied to the present invention, so as to achieve the purpose of the present invention.
  • the classification model in the present invention adopts a neural network model, which originated from an algorithm that attempts to make a machine imitate the brain, and connects the cranial nerve units imitating neurons to form a network-like graph.
  • the classification model can be processed into building blocks to obtain graphical classification building blocks.
  • the graphical classification building block includes an input layer, a hidden layer and an output layer;
  • the input layer is used to input feature values of things to be predicted.
  • the characteristic value of wine is the color and alcohol concentration of the wine; in disease judgment, the characteristic value of disease becomes laboratory data.
  • the hidden layer is used to extract salient features in the input data, classify according to the salient features, and transmit the classification results to the output layer;
  • the hidden layer can be a single layer or multiple layers, each hidden There are corresponding numbers of neurons in each layer, and the number of neurons in each layer can be the same or different.
  • the output layer is used to output classification results.
  • the output classification results are beer, red wine, and white wine
  • the output classification results are health, cold, fever, etc.
  • the entire neural network collects information through the input layer, uses the hidden layer to perform operations and processing information, and then outputs the classification results from the output layer, thus achieving the function of a classifier.
  • the graphical classification building block includes 4 input layers, 2 hidden layers (including 4 hidden units and 3 hidden units), and 2 outputs Floor.
  • the hidden unit is the neuron.
  • step S2 training data is used to train the graphical classification building block.
  • Whether the prediction result (the output classification result) is accurate is related to the structure of the neural network, the number of training times, and even some random factors during training.
  • extract the uploaded excel data convert the excel into matrix data, and use the matrix data as training data.
  • the input layer is the training data feature, and the building blocks are shown in Figure 3.
  • the neural network input layer is 2;
  • the neural network hidden layer structure uses building blocks As shown in Figure 4; after setting here, the hidden layer structure of the neural network is 2 layers, and each layer is 4 structural units.
  • the output layer structure of the neural network uses building blocks as shown in Figure 5. Because in the data, there are only two classification results of Iris mountain and Iris variegated, so after setting here, the output layer is 2.
  • the number of training times should reach a certain number, for example, up to 400 times, so that the classification result can be more accurate.
  • Backpropagation algorithm the full name BackPropagation algorithm, is a learning algorithm suitable for multi-layer neural networks under guidance. It is based on the gradient descent method. In the present invention, all are implemented by front-end code. Below, briefly introduce the principle of the following backward propagation algorithm:
  • the model Before training, the model is initialized randomly. If this model is used for prediction at this time, even if the data given is training data with results, the results obtained are also random.
  • each set of correct training data is sent to the neural network to obtain the corresponding excitation response, and then the excitation corresponding to the target output corresponding to the training input is calculated to obtain the hidden layer and the output layer The corresponding error. Then, multiply the input stimulus and response error to obtain the gradient of the weight, and then multiply this gradient by a constant "training therefore" and take the inverse and add it to the weight.
  • the error backward propagation method is completed, and the updated weights are recalculated, and the continuous iteration is the process of continuous training of the model.
  • the model After the model is trained, let the model predict the training data at this time. Through the new weight calculation, the model will accurately predict the classification result of the training data.
  • the classification model can be divided into red (bottom right) and blue (top left) according to the feature value. At this time, the training has been successful.
  • the trained model can be kept in the file, and the model can be used.
  • the feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
  • the model can be used to predict the new input feature values, where the number of input feature values needs to match the number of feature values in the training data. Otherwise, accurate prediction cannot be made.
  • the prediction process is to calculate the weight of the model with the newly input feature values to obtain the classification result predicted under the model.
  • the "predicted classification result” building block the output is the type of classification, here is one of the mountain iris and the variegated iris.
  • “predicted classification probability” the output is a list type, that is, the probability of iris and variegated respectively is predicted.
  • the invention realizes the classification AI function in the graphical programming tool.
  • the graphical classification building blocks are used to realize the classifier function based on the feature value of the data class, and its technical methods are packaged as graphical classification building blocks, which can be directly used in graphical programming tools.
  • users can edit programming works with classified AI, expand the gameplay and program types of the written program, and improve the advanced nature of programming tools in the field of children's programming. It can be applied to high school artificial intelligence compulsory courses. Course area.
  • the present invention also provides an electronic device 10, as shown in FIG. 7, which includes:
  • the processor 110 is adapted to implement various instructions, and
  • the storage device 120 is adapted to store multiple instructions, and the instructions are adapted to be loaded and executed by the processor 110:
  • the feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
  • the processor 110 may be a general-purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine) or other programmable logic device, Discrete gate or transistor logic, discrete hardware components, or any combination of these components.
  • the processor may also be any conventional processor, microprocessor or state machine.
  • the processor may also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, any other such configuration.
  • the storage device 120 can be used to store non-volatile software programs, non-volatile computer-executable programs and modules, such as those based on graphical programming tools in the embodiments of the present invention.
  • the processor executes various functional applications and data processing of the classification AI implementation method based on graphical programming tools by running the non-volatile software programs, instructions, and units stored in the storage device, thus realizing the foregoing method embodiments.
  • the present invention also provides a non-volatile computer-readable storage medium that stores computer-executable instructions.
  • the computer-executable instructions are executed by one or more processors, they can The one or more processors are caused to execute the classification AI implementation method based on the graphical programming tool.
  • the present invention also provides a computer program product.
  • the computer program product includes a computer program stored on a non-volatile computer-readable storage medium.
  • the computer program includes program instructions. When the program instructions are executed by a processor, the processor Perform the described classification AI implementation method based on graphical programming tools.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Stored Programmes (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

Disclosed in the present invention are an AI implementation method for classification based on a graphical programming tool, and an electronic device. Said method comprises the steps of: creating a classification model, and performing modularization processing on the classification model to obtain graphical classification modules; training the graphical classification modules by using training data; and inputting characteristic values of an object to be predicted in the graphical classification modules, analyzing the characteristic values of the object to be predicted by means of the graphical classification modules, and outputting a classification result of the object to be predicted. The present invention implements a data class characteristic value-based classifier function by using graphical classification modules, and packages the data class characteristic values into graphical modules, which can be directly used for a graphical programming tool. By using the graphical modules, a user can compile programming works with classification AI, and the game playing methods and program types of compiled programs are expanded.

Description

一种基于图形化编程工具的分类AI实现方法及电子设备A classification AI realization method and electronic equipment based on graphical programming tools 技术领域Technical field
本发明涉及计算机技术领域，尤其涉及一种基于图形化编程工具的分类AI实现方法及电子设备。The present invention relates to the field of computer technology, in particular to a classification AI implementation method and electronic equipment based on graphical programming tools.
背景技术Background technique
随着社会的发展和进步，以及计算机科学的飞速发展，编程成为了现代人一项越来越重要的基本能力，而不仅仅是IT行业人员的专职工作。任何年龄、任何基础的人员都有越来越强的学习和体验程序编写的需求，图形化编程工具随着时代的发展应运而生。使用图形化编程工具，用户仅仅通过拖拽编程组件，就能独立完成具有复杂逻辑的程序编写。With the development and progress of society and the rapid development of computer science, programming has become an increasingly important basic ability for modern people, not just a full-time job for IT industry personnel. People of any age and any foundation have increasingly strong demands for learning and experience programming. Graphical programming tools have emerged with the development of the times. Using graphical programming tools, users can independently complete programming with complex logic just by dragging and dropping programming components.
但是在现在的图形化编程工具中，很少涉及到人工智能领域。通常，就算与人工智能相关，也仅仅是调用其他公司公开的API，仅能实现部分简单功能，具有很大的局限性。这一局限性使得这类图形化编程工具的应用只能停留在编程逻辑教育，少儿steam教育等领域，无法进一步扩展，例如不具有智能分类功能。But in the current graphical programming tools, the field of artificial intelligence is rarely involved. Usually, even if it is related to artificial intelligence, it is only to call other companies' public APIs, which can only implement some simple functions, which has great limitations. This limitation makes the application of this kind of graphical programming tools can only stay in the fields of programming logic education, children's steam education, etc., and cannot be further expanded. For example, they do not have intelligent classification functions.
因此，现有技术还有待于改进和发展。Therefore, the existing technology needs to be improved and developed.
发明内容Summary of the invention
鉴于上述现有技术的不足，本发明的目的在于提供一种基于图形化编程工具的分类AI实现方法及电子设备，旨在解决现有编程方法不具有智能分类功能的问题。In view of the foregoing shortcomings of the prior art, the purpose of the present invention is to provide a classification AI implementation method and electronic equipment based on a graphical programming tool, aiming to solve the problem that the existing programming method does not have an intelligent classification function.
本发明的技术方案如下：The technical scheme of the present invention is as follows:
一种基于图形化编程工具的分类AI实现方法，其中，包括步骤：A classification AI implementation method based on graphical programming tools, including the following steps:
创建分类模型，并对所述分类模型进行积木化处理，得到图形化分类积木；Creating a classification model, and performing building block processing on the classification model to obtain a graphical classification building block;
使用训练数据对所述图形化分类积木进行训练；Use training data to train the graphical classification building block;
在所述图形化分类积木中输入待预测事物的特征值，通过所述图形化分类积木对待预测事物的特征值进行分析，输出待预测事物的分类结果。The feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
所述的基于图形化编程工具的分类AI实现方法，其中，使用后向传播算法对所述图形化分类积木进行训练。In the method for implementing classification AI based on graphical programming tools, a backward propagation algorithm is used to train the graphical classification building blocks.
所述的基于图形化编程工具的分类AI实现方法，其中，所述图形化分类积木包括输入层、隐藏层和输出层；In the method for implementing classification AI based on graphical programming tools, wherein the graphical classification building blocks include an input layer, a hidden layer, and an output layer;
其中，所述输入层用于输入待预测事物的特征值；Wherein, the input layer is used to input feature values of things to be predicted;
所述隐藏层用于提取输入数据中的显著特征，并根据所述显著特征进行分类，以及将分类结果传输到输出层；The hidden layer is used to extract salient features in the input data, classify them according to the salient features, and transmit the classification results to the output layer;
所述输出层用于输出分类结果。The output layer is used to output classification results.
所述的基于图形化编程工具的分类AI实现方法，其中，待预测事物的特征值与训练数据中的特征值数量匹配。In the method for implementing classification AI based on graphical programming tools, the feature value of the thing to be predicted matches the number of feature values in the training data.
所述的基于图形化编程工具的分类AI实现方法，其中，所述隐藏层为一层或多层，且每层隐藏层均具有相应数量的神经元。In the method for implementing classified AI based on graphical programming tools, the hidden layer is one or more layers, and each hidden layer has a corresponding number of neurons.
所述的基于图形化编程工具的分类AI实现方法，其中，每一组训练数据均包括特征值和分类结果。In the method for implementing classification AI based on graphical programming tools, each set of training data includes feature values and classification results.
所述的基于图形化编程工具的分类AI实现方法，其中，提取上传的excel数据，将所述excel转换为矩阵数据，并所述矩阵数据作为训练数据。In the method for implementing classification AI based on graphical programming tools, the uploaded excel data is extracted, the excel is converted into matrix data, and the matrix data is used as training data.
一种电子设备，其中，包括：An electronic device, which includes:
处理器，适于实现各指令，以及A processor suitable for implementing instructions, and
存储设备，适于存储多条指令，所述指令适于由处理器加载并执行：The storage device is adapted to store multiple instructions, and the instructions are adapted to be loaded and executed by the processor:
创建分类模型，并对所述分类模型进行积木化处理，得到图形化分类积木；Creating a classification model, and performing building block processing on the classification model to obtain a graphical classification building block;
使用训练数据对所述图形化分类积木进行训练；Use training data to train the graphical classification building block;
在所述图形化分类积木中输入待预测事物的特征值，通过所述图形化分类积木对待预测事物的特征值进行分析，输出待预测事物的分类结果。The feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
一种非易失性计算机可读存储介质，其中，所述非易失性计算机可读存储介质存储有计算机可执行指令，该计算机可执行指令被一个或多个处理器执行时，可使得所述一个或多个处理器执行所述的基于图形化编程工具的分类AI实现方法。A non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores computer-executable instructions that, when executed by one or more processors, can cause all The one or more processors execute the classification AI implementation method based on the graphical programming tool.
一种计算机程序产品，其中，计算机程序产品包括存储在非易失性计算机可读存储介质上的计算机程序，计算机程序包括程序指令，当程序指令被处理器执行时，使所述处理器执行所述的基于图形化编程工具的分类AI实现方法。A computer program product, wherein the computer program product includes a computer program stored on a non-volatile computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a processor, the processor executes all The described classification AI implementation method based on graphical programming tools.
有益效果：本发明利用图形化分类积木，实现了基于数据类特征值的分类器功能，并将其包装为图形化积木，可直接用于图形化编程工具。通过运用这些图形化积木，用户可编辑具备分类AI的编程作品，扩展了编写出程序的游戏玩法、程序类型。Beneficial effects: The present invention uses graphical classification building blocks to realize the function of a classifier based on the characteristic values of the data type, and packages them into graphical building blocks, which can be directly used in graphical programming tools. By using these graphical building blocks, users can edit programming works with classified AI, which expands the gameplay and program types of the programmed programs.
附图说明Description of the drawings
图1为本发明一种基于图形化编程工具的分类AI实现方法较佳实施例的流程图。FIG. 1 is a flowchart of a preferred embodiment of a classification AI implementation method based on a graphical programming tool of the present invention.
图2为本发明中一个神经网络实例的网络状图。Figure 2 is a network diagram of an example of a neural network in the present invention.
图3为本发明中一个神经网络输入层实例的积木图。Fig. 3 is a block diagram of an example of a neural network input layer in the present invention.
图4为本发明中一个神经网络隐含层实例的积木图。Figure 4 is a block diagram of an example of a hidden layer of a neural network in the present invention.
图5为本发明中一个神经网络输出层实例的积木图。Fig. 5 is a block diagram of an example of a neural network output layer in the present invention.
图6为本发明中一个神经网络的训练结果示意图。Fig. 6 is a schematic diagram of the training result of a neural network in the present invention.
图7为本发明一种电子设备较佳实施例的结构框图。FIG. 7 is a structural block diagram of a preferred embodiment of an electronic device of the present invention.
具体实施方式detailed description
本发明提供一种基于图形化编程工具的分类AI实现方法及电子设备，为使本发明的目的、技术方案及效果更加清楚、明确，以下对本发明进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。The present invention provides a classification AI implementation method and electronic equipment based on a graphical programming tool. In order to make the objectives, technical solutions and effects of the present invention clearer and clearer, the present invention will be described in further detail below. It should be understood that the specific embodiments described here are only used to explain the present invention, but not to limit the present invention.
请参阅图1，图1为本发明一种基于图形化编程工具的分类AI实现方法较佳实施例的流程图，如图所示，其包括步骤：Please refer to FIG. 1. FIG. 1 is a flowchart of a preferred embodiment of a classification AI implementation method based on a graphical programming tool of the present invention. As shown in the figure, it includes the steps:
S1、创建分类模型，并对所述分类模型进行积木化处理，得到图形化分类积木；S1. Create a classification model, and perform building block processing on the classification model to obtain a graphical classification building block;
S2、使用训练数据对所述图形化分类积木进行训练；S2. Use training data to train the graphical classification building block;
S3、在所述图形化分类积木中输入待预测事物的特征值，通过所述图形化分类积木对待预测事物的特征值进行分析，输出待预测事物的分类结果。S3. Input the feature value of the object to be predicted into the graphical classification building block, analyze the feature value of the object to be predicted through the graphical classification building block, and output the classification result of the object to be predicted.
本发明利用图形化分类积木，实现了基于数据类特征值的分类器功能，并将其包装为图形化积木，可直接用于图形化编程工具。通过运用这些图形化积木，用户可编辑具备分类AI的编程作品，扩展了编写出程序的游戏玩法、程序类型。The invention utilizes graphical classification building blocks to realize the function of a classifier based on the characteristic value of the data type, and packs them into graphical building blocks, which can be directly used in graphical programming tools. By using these graphical building blocks, users can edit programming works with classified AI, which expands the gameplay and program types of the programmed programs.
具体地，在步骤S1中，先创建分类模型。Specifically, in step S1, a classification model is created first.
其中的分类模型类似于分类器。分类器是数据挖掘中对样本进行分类的方法的统称。The classification model is similar to a classifier. Classifier is a general term for the method of classifying samples in data mining.
举例来说，对于酒类识别：可测量一瓶酒的颜色(波长)、酒精浓度等特征值，机器根据这些特征值判断这瓶酒是属于啤酒、红酒、白酒中的哪一种。疾病判断：病人到医院检测了肝功、血液测验等数据，将这些测验数据送进一个机器里，机器根据这些数据来判断这个病人是否得病，得的什么病。这种能够对输入的内容进行自动分类的机器，即称为分类器。For example, for wine identification: the color (wavelength), alcohol concentration and other characteristic values of a bottle of wine can be measured, and the machine judges which one of beer, red wine, and white wine the bottle belongs to based on these characteristic values. Disease judgment: The patient goes to the hospital to check the liver function, blood test and other data. These test data are sent to a machine, and the machine judges whether the patient is sick or not based on the data. This kind of machine that can automatically classify the input is called a classifier.
对人类来说，单凭肉眼可能也能对酒类进行一个简单识别，但却无法 对疾病进行判断，这个时候，如果提供一个疾病判断的分类器，那么只需要输入相应的检测数据，就可以得出判断结果。For humans, the naked eye may be able to make a simple identification of alcohol, but it is unable to judge the disease. At this time, if a classifier for disease judgment is provided, then only the corresponding detection data needs to be input. Get the judgment result.
而基于人工智能的分类器不仅仅能完成疾病判断这种程度的事情，其可以完成更复杂、更多特征值的分类判断。实现分类器有很多方法，基于神经网络的深度学习就是其中一种。本发明也主要是采用神经网络的深度学习来实现分类器，但显然对于其他类型的分类器，同样可以应用在本发明中，从而实现本发明的目的。And the classifier based on artificial intelligence can not only complete disease judgments to this degree, it can complete more complicated classification judgments with more feature values. There are many ways to implement a classifier, and deep learning based on neural networks is one of them. The present invention also mainly adopts the deep learning of neural network to realize the classifier, but obviously for other types of classifiers, it can also be applied to the present invention, so as to achieve the purpose of the present invention.
本发明中的分类模型采用神经网络模型，其起源于尝试让机器模仿大脑的算法，将模仿神经元的脑神经单位进行连接，形成网络状图。The classification model in the present invention adopts a neural network model, which originated from an algorithm that attempts to make a machine imitate the brain, and connects the cranial nerve units imitating neurons to form a network-like graph.
具体地，可对所述分类模型进行积木化处理，得到图形化分类积木。Specifically, the classification model can be processed into building blocks to obtain graphical classification building blocks.
进一步，所述图形化分类积木包括输入层、隐藏层和输出层；Further, the graphical classification building block includes an input layer, a hidden layer and an output layer;
其中，所述输入层用于输入待预测事物的特征值。比如：酒类识别中，酒的特征值就是酒的颜色和酒精浓度；疾病判断中，疾病的特征值就变成了化验数据。Wherein, the input layer is used to input feature values of things to be predicted. For example, in wine recognition, the characteristic value of wine is the color and alcohol concentration of the wine; in disease judgment, the characteristic value of disease becomes laboratory data.
所述隐藏层用于提取输入数据中的显著特征，并根据所述显著特征进行分类，以及将分类结果传输到输出层；隐藏层可以是单层的，也可以是多层的，每一隐藏层里都有相应数量的神经元，并且每一层中的神经元数量可以相同、也可以不同。The hidden layer is used to extract salient features in the input data, classify according to the salient features, and transmit the classification results to the output layer; the hidden layer can be a single layer or multiple layers, each hidden There are corresponding numbers of neurons in each layer, and the number of neurons in each layer can be the same or different.
所述输出层用于输出分类结果。比如：酒类识别中，输出的分类结果是啤酒、红酒、白酒；疾病判断中，输出的分类结果是健康、感冒、发烧等。The output layer is used to output classification results. For example: in wine recognition, the output classification results are beer, red wine, and white wine; in disease judgment, the output classification results are health, cold, fever, etc.
总的来说，整个神经网络通过输入层收集信息，利用隐藏层进行运算和处理信息，再从输出层输出分类结果，这样就做到了分类器的功能。In general, the entire neural network collects information through the input layer, uses the hidden layer to perform operations and processing information, and then outputs the classification results from the output layer, thus achieving the function of a classifier.
本发明中，由于将分类模型进行了积木化处理，便得到了图形化分类积木。In the present invention, since the classification model is processed into building blocks, a graphical classification building block is obtained.
如图2所示，在一个具体实例中，所述的图形化分类积木(神经网络) 包括4个输入层、2个隐藏层(分别包括4个隐藏单元和3个隐藏单元)、2个输出层。其中的隐藏单元即为神经元。As shown in Figure 2, in a specific example, the graphical classification building block (neural network) includes 4 input layers, 2 hidden layers (including 4 hidden units and 3 hidden units), and 2 outputs Floor. The hidden unit is the neuron.
分类器需要通过训练来大概地预测出分类结果。所以在所述步骤S2中，使用训练数据对所述图形化分类积木进行训练。The classifier needs to be trained to roughly predict the classification result. Therefore, in the step S2, training data is used to train the graphical classification building block.
举例说明：一个从来没见过酒的人和一个对酒很有研究的人，肯定是后者更能分辨出酒的种类，对酒很有研究的人，非常了解酒的种类和味道，喝酒的次数越多，知道的酒的品种也就越多。For example: a person who has never seen wine and a person who is very researched about wine, it must be the latter who can better distinguish the types of wine. People who are very researched about wine know the types and tastes of wine very well. The more you use it, the more varieties you know.
同样的，对于分类器来说，它也需要大量的训练数据不断进行训练，并且每一组的训练数据需要包含特征值以及分类结果。类似于训练一个不懂酒的人，训练的次数多了，也能分辨出酒的种类。Similarly, for a classifier, it also requires a large amount of training data for continuous training, and each group of training data needs to include feature values and classification results. It's similar to training a person who doesn't understand alcohol. If you train more often, you can distinguish the type of alcohol.
预测结果(输出的分类结果)是否准确，与神经网络的结构、训练次数、甚至训练时的一些随机发生的因素都有关系。Whether the prediction result (the output classification result) is accurate is related to the structure of the neural network, the number of training times, and even some random factors during training.
进一步，提取上传的excel数据，将所述excel转换为矩阵数据，并所述矩阵数据作为训练数据。Further, extract the uploaded excel data, convert the excel into matrix data, and use the matrix data as training data.
用户可自行上传相关训练数据的excel表格，***会存作矩阵数据，即二维数组。Users can upload the excel form of relevant training data by themselves, and the system will save it as matrix data, that is, a two-dimensional array.
例如点击“上传excel为矩阵按钮”，选择对应的excel文件即可完成训练数据的上传。上传后，可以通过矩阵积木，将矩阵取为列表运用。For example, click "upload excel as matrix button" and select the corresponding excel file to complete the upload of training data. After uploading, you can use the matrix building block to take the matrix as a list.
在一个具体实例中，在图形化编程工具中，输入层则是训练数据特征，使用积木如图3所示，此处设置后，神经网络输入层为2；神经网络隐含层结构，使用积木如图4所示；此处设置后，神经网络隐含层结构，为2层，每层分别是4个结构单元，神经网络输出层结构，使用积木如图5所示。因为在数据中，仅有山鸢尾和杂色鸢尾两种分类结果，因此此处设置后，输出层为2。In a specific example, in the graphical programming tool, the input layer is the training data feature, and the building blocks are shown in Figure 3. After setting here, the neural network input layer is 2; the neural network hidden layer structure uses building blocks As shown in Figure 4; after setting here, the hidden layer structure of the neural network is 2 layers, and each layer is 4 structural units. The output layer structure of the neural network uses building blocks as shown in Figure 5. Because in the data, there are only two classification results of Iris mountain and Iris variegated, so after setting here, the output layer is 2.
进一步，使用后向传播算法对所述图形化分类积木进行训练，训练次数应达到一定数量，例如达到400次，这样可以使分类结果更加准确。Further, using the backward propagation algorithm to train the graphical classification building blocks, the number of training times should reach a certain number, for example, up to 400 times, so that the classification result can be more accurate.
后向传播算法，全称BackPropagation算法，是在有指导下，适合于多层神经元网络的一种学习算法，它建立在梯度下降法的基础上，在本发明中，均由前端代码实现。下面，简单介绍以下后向传播算法的原理：Backpropagation algorithm, the full name BackPropagation algorithm, is a learning algorithm suitable for multi-layer neural networks under guidance. It is based on the gradient descent method. In the present invention, all are implemented by front-end code. Below, briefly introduce the principle of the following backward propagation algorithm:
在训练以前，模型是初始化随机的，这时如果用此模型进行预测，就算给到的数据是有结果的训练数据，得到的结果也是随机的。在后向传播算法中，将每一组正确的训练数据，送入神经网络，以获得相应的激励响应，再将激励相应同训练输入对应的目标输出求差，从而获得隐含层和输出层的相应误差。然后，将输入激励和响应误差相乘，获得权重的梯度，再将这个梯度乘上一个常数“训练因此”并取反后加在权重上。这样误差后向传播法就完成了，再把更新的权重重新计算，不停的迭代，就是模型在不断训练的过程。当模型训练好后，此时再让模型去预测训练数据，经由新的权重计算，模型将会准确预测出训练数据的分类结果。Before training, the model is initialized randomly. If this model is used for prediction at this time, even if the data given is training data with results, the results obtained are also random. In the backward propagation algorithm, each set of correct training data is sent to the neural network to obtain the corresponding excitation response, and then the excitation corresponding to the target output corresponding to the training input is calculated to obtain the hidden layer and the output layer The corresponding error. Then, multiply the input stimulus and response error to obtain the gradient of the weight, and then multiply this gradient by a constant "training therefore" and take the inverse and add it to the weight. In this way, the error backward propagation method is completed, and the updated weights are recalculated, and the continuous iteration is the process of continuous training of the model. After the model is trained, let the model predict the training data at this time. Through the new weight calculation, the model will accurately predict the classification result of the training data.
如图6所示，经过多次的训练，分类模型已经可以根据特征值，分成了红色(右下)和蓝色(左上)两类，此时训练已经成功。训练后的模型可保留在文件中，可将模型加以运用。As shown in Figure 6, after many trainings, the classification model can be divided into red (bottom right) and blue (top left) according to the feature value. At this time, the training has been successful. The trained model can be kept in the file, and the model can be used.
在所述步骤S3中，在所述图形化分类积木中输入待预测事物的特征值，通过所述图形化分类积木对待预测事物的特征值进行分析，输出待预测事物的分类结果。In the step S3, the feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
也就是说，分类模型(也就是图形化分类积木)训练完毕后，就可以使用模型，对新输入的特征值进行预测，其中，输入的特征值数量需要与训练数据中的特征值数量匹配，否则无法进行准确预测。In other words, after the classification model (that is, the graphical classification building block) is trained, the model can be used to predict the new input feature values, where the number of input feature values needs to match the number of feature values in the training data. Otherwise, accurate prediction cannot be made.
预测的过程，就是用新输入的特征值对模型的权重进行计算，得到在该模型下预测的分类结果。The prediction process is to calculate the weight of the model with the newly input feature values to obtain the classification result predicted under the model.
“预测分类结果”积木，输出为分类的种类，在这里就是山鸢尾和杂色鸢尾其中一种。而“预测分类概率”，输出为列表类型，即为分别预测是山鸢尾和杂色鸢尾的概率。The "predicted classification result" building block, the output is the type of classification, here is one of the mountain iris and the variegated iris. And "predicted classification probability", the output is a list type, that is, the probability of iris and variegated respectively is predicted.
本发明在图形化编程工具中实现了分类AI功能。利用图形化分类积木，实现了基于数据类特征值的分类器功能，并将其技术方法包装为图形化分类积木，可直接用于图形化编程工具。通过运用这些图形化分类积木，用户可编辑具备分类AI的编程作品，扩展了编写出程序的游戏玩法、程序类型，同时提升了编程工具在少儿编程领域的先进性，可以应用于高中人工智能必修课领域。The invention realizes the classification AI function in the graphical programming tool. The graphical classification building blocks are used to realize the classifier function based on the feature value of the data class, and its technical methods are packaged as graphical classification building blocks, which can be directly used in graphical programming tools. Through the use of these graphical classification building blocks, users can edit programming works with classified AI, expand the gameplay and program types of the written program, and improve the advanced nature of programming tools in the field of children's programming. It can be applied to high school artificial intelligence compulsory courses. Course area.
本发明还提供一种电子设备10，如图7所示，其包括：The present invention also provides an electronic device 10, as shown in FIG. 7, which includes:
处理器110，适于实现各指令，以及The processor 110 is adapted to implement various instructions, and
存储设备120，适于存储多条指令，所述指令适于由处理器110加载并执行：The storage device 120 is adapted to store multiple instructions, and the instructions are adapted to be loaded and executed by the processor 110:
创建分类模型，并对所述分类模型进行积木化处理，得到图形化分类积木；Creating a classification model, and performing building block processing on the classification model to obtain a graphical classification building block;
使用训练数据对所述图形化分类积木进行训练；Use training data to train the graphical classification building block;
在所述图形化分类积木中输入待预测事物的特征值，通过所述图形化分类积木对待预测事物的特征值进行分析，输出待预测事物的分类结果。The feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
所述处理器110可以为通用处理器、数字信号处理器(DSP)、专用集成电路(ASIC)、现场可编程门阵列(FPGA)、单片机、ARM(Acorn RISC Machine)或其它可编程逻辑器件、分立门或晶体管逻辑、分立的硬件组件或者这些部件的任何组合。还有，处理器还可以是任何传统处理器、微处理器或状态机。处理器也可以被实现为计算设备的组合，例如，DSP和微处理器的组合、多个微处理器、一个或多个微处理器结合DSP核、任何其它这种配置。The processor 110 may be a general-purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine) or other programmable logic device, Discrete gate or transistor logic, discrete hardware components, or any combination of these components. In addition, the processor may also be any conventional processor, microprocessor or state machine. The processor may also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, any other such configuration.
存储设备120作为一种非易失性计算机可读存储介质，可用于存储非易失性软件程序、非易失性计算机可执行程序以及模块，如本发明实施例中的基于图形化编程工具的分类AI实现方法对应的程序指令。处理器通过运行存储在存储设备中的非易失性软件程序、指令以及单元，从而执行基 于图形化编程工具的分类AI实现方法的各种功能应用以及数据处理，即实现上述方法实施例。As a non-volatile computer-readable storage medium, the storage device 120 can be used to store non-volatile software programs, non-volatile computer-executable programs and modules, such as those based on graphical programming tools in the embodiments of the present invention. Classify the program instructions corresponding to the AI implementation method. The processor executes various functional applications and data processing of the classification AI implementation method based on graphical programming tools by running the non-volatile software programs, instructions, and units stored in the storage device, thus realizing the foregoing method embodiments.
关于上述电子设备10的具体技术细节在前面的方法中已有详述，故不再赘述。The specific technical details of the above-mentioned electronic device 10 have been described in detail in the previous method, so it will not be repeated.
本发明还提供一种非易失性计算机可读存储介质，所述非易失性计算机可读存储介质存储有计算机可执行指令，该计算机可执行指令被一个或多个处理器执行时，可使得所述一个或多个处理器执行所述的基于图形化编程工具的分类AI实现方法。The present invention also provides a non-volatile computer-readable storage medium that stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, they can The one or more processors are caused to execute the classification AI implementation method based on the graphical programming tool.
本发明还提供一种计算机程序产品，计算机程序产品包括存储在非易失性计算机可读存储介质上的计算机程序，计算机程序包括程序指令，当程序指令被处理器执行时，使所述处理器执行所述的基于图形化编程工具的分类AI实现方法。The present invention also provides a computer program product. The computer program product includes a computer program stored on a non-volatile computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a processor, the processor Perform the described classification AI implementation method based on graphical programming tools.
应当理解的是，本发明的应用不限于上述的举例，对本领域普通技术人员来说，可以根据上述说明加以改进或变换，所有这些改进和变换都应属于本发明所附权利要求的保护范围。It should be understood that the application of the present invention is not limited to the above examples. For those of ordinary skill in the art, improvements or changes can be made based on the above description, and all these improvements and changes should fall within the protection scope of the appended claims of the present invention.

Claims (10)

  1. 一种基于图形化编程工具的分类AI实现方法，其特征在于，包括步骤：A method for implementing classified AI based on graphical programming tools is characterized in that it comprises the following steps:
    创建分类模型，并对所述分类模型进行积木化处理，得到图形化分类积木；Creating a classification model, and performing building block processing on the classification model to obtain a graphical classification building block;
    使用训练数据对所述图形化分类积木进行训练；Use training data to train the graphical classification building block;
    在所述图形化分类积木中输入待预测事物的特征值，通过所述图形化分类积木对待预测事物的特征值进行分析，输出待预测事物的分类结果。The feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
  2. 根据权利要求1所述的基于图形化编程工具的分类AI实现方法，其特征在于，使用后向传播算法对所述图形化分类积木进行训练。The method for implementing classification AI based on graphical programming tools according to claim 1, wherein a backward propagation algorithm is used to train the graphical classification building blocks.
  3. 根据权利要求1所述的基于图形化编程工具的分类AI实现方法，其特征在于，所述图形化分类积木包括输入层、隐藏层和输出层；The method for implementing classification AI based on graphical programming tools according to claim 1, wherein the graphical classification building blocks include an input layer, a hidden layer and an output layer;
    其中，所述输入层用于输入待预测事物的特征值；Wherein, the input layer is used to input feature values of things to be predicted;
    所述隐藏层用于提取输入数据中的显著特征，并根据所述显著特征进行分类，以及将分类结果传输到输出层；The hidden layer is used to extract salient features in the input data, classify them according to the salient features, and transmit the classification results to the output layer;
    所述输出层用于输出分类结果。The output layer is used to output classification results.
  4. 根据权利要求1所述的基于图形化编程工具的分类AI实现方法，其特征在于，待预测事物的特征值与训练数据中的特征值数量匹配。The method for implementing classification AI based on graphical programming tools according to claim 1, wherein the feature value of the thing to be predicted matches the number of feature values in the training data.
  5. 根据权利要求3所述的基于图形化编程工具的分类AI实现方法，其特征在于，所述隐藏层为一层或多层，且每层隐藏层均具有相应数量的神经元。The method for implementing classification AI based on graphical programming tools according to claim 3, wherein the hidden layer is one or more layers, and each hidden layer has a corresponding number of neurons.
  6. 根据权利要求1所述的基于图形化编程工具的分类AI实现方法，其特征在于，每一组训练数据均包括特征值和分类结果。The method for implementing classification AI based on graphical programming tools according to claim 1, wherein each set of training data includes feature values and classification results.
  7. 根据权利要求6所述的基于图形化编程工具的分类AI实现方法，其特征在于，提取上传的excel数据，将所述excel转换为矩阵数据，并所述矩阵数据作为训练数据。The method for implementing classification AI based on graphical programming tools according to claim 6, wherein the uploaded excel data is extracted, the excel is converted into matrix data, and the matrix data is used as training data.
  8. 一种电子设备，其特征在于，包括：An electronic device, characterized in that it comprises:
    处理器，适于实现各指令，以及A processor suitable for implementing instructions, and
    存储设备，适于存储多条指令，所述指令适于由处理器加载并执行：The storage device is adapted to store multiple instructions, and the instructions are adapted to be loaded and executed by the processor:
    创建分类模型，并对所述分类模型进行积木化处理，得到图形化分类积木；Creating a classification model, and performing building block processing on the classification model to obtain a graphical classification building block;
    使用训练数据对所述图形化分类积木进行训练；Use training data to train the graphical classification building block;
    在所述图形化分类积木中输入待预测事物的特征值，通过所述图形化分类积木对待预测事物的特征值进行分析，输出待预测事物的分类结果。The feature value of the object to be predicted is input into the graphical classification building block, the feature value of the object to be predicted is analyzed through the graphical classification building block, and the classification result of the object to be predicted is output.
  9. 一种非易失性计算机可读存储介质，其特征在于，所述非易失性计算机可读存储介质存储有计算机可执行指令，该计算机可执行指令被一个或多个处理器执行时，可使得所述一个或多个处理器执行权利要求1-7任一项所述的基于图形化编程工具的分类AI实现方法。A non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, The one or more processors are caused to execute the classification AI implementation method based on a graphical programming tool according to any one of claims 1-7.
  10. 一种计算机程序产品，其特征在于，计算机程序产品包括存储在非易失性计算机可读存储介质上的计算机程序，计算机程序包括程序指令，当程序指令被处理器执行时，使所述处理器执行权利要求1-7任一项所述的基于图形化编程工具的分类AI实现方法。A computer program product, characterized in that the computer program product includes a computer program stored on a non-volatile computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by a processor, the processor Implement the classification AI implementation method based on graphical programming tools according to any one of claims 1-7.
PCT/CN2019/100384 2019-02-19 2019-08-13 Ai implementation method for classification based on graphical programming tool, and electronic device WO2020168690A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910121926.9 2019-02-19
CN201910121926.9A CN109978000A (en) 2019-02-19 2019-02-19 A kind of classification AI implementation method and electronic equipment based on graphic programming tool

Publications (1)

Publication Number Publication Date
WO2020168690A1 true WO2020168690A1 (en) 2020-08-27

Family

ID=67077018

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/100384 WO2020168690A1 (en) 2019-02-19 2019-08-13 Ai implementation method for classification based on graphical programming tool, and electronic device

Country Status (2)

Country Link
CN (1) CN109978000A (en)
WO (1) WO2020168690A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113033297A (en) * 2021-02-08 2021-06-25 深圳市优必选科技股份有限公司 Object programming method, device, equipment and storage medium

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109978000A (en) * 2019-02-19 2019-07-05 深圳点猫科技有限公司 A kind of classification AI implementation method and electronic equipment based on graphic programming tool
CN113966494A (en) * 2019-08-27 2022-01-21 西门子股份公司 System, method and storage medium for supporting graphical programming based on neuron blocks
CN111522542B (en) * 2020-03-16 2023-10-27 北京编程猫科技有限公司 Programming method and device for perfecting graphical programming building block tool
CN111841017A (en) * 2020-05-29 2020-10-30 北京编程猫科技有限公司 Game AI programming realization method and device
CN115223424A (en) * 2022-03-23 2022-10-21 北京好奇星教育科技有限公司 Programming method and device and electronic equipment

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104503754A (en) * 2014-12-16 2015-04-08 江南大学 Programming and compiling design method in robot graphical programming system
CN107169506A (en) * 2017-04-14 2017-09-15 微梦创科网络科技（中国）有限公司 Random assortment method and device based on assembled classifier
CN207233274U (en) * 2017-07-06 2018-04-13 深圳市魔块智能有限公司 Wireless programmable electronic modular system
CN108460399A (en) * 2017-12-29 2018-08-28 华南师范大学 A kind of child building block builds householder method and system
CN108984693A (en) * 2018-07-04 2018-12-11 深圳点猫科技有限公司 A kind of sharing method and system of the program based on artificial intelligence
CN109284100A (en) * 2018-07-30 2019-01-29 深圳点猫科技有限公司 The rendering method and system of graphical programming language for subject cognitive tool
US20190042938A1 (en) * 2017-05-19 2019-02-07 Tata Consultancy Services Limited Simultaneous multi-class learning for data classification
CN109344959A (en) * 2018-08-27 2019-02-15 联想(北京)有限公司 Neural network training method, nerve network system and computer system
CN109978000A (en) * 2019-02-19 2019-07-05 深圳点猫科技有限公司 A kind of classification AI implementation method and electronic equipment based on graphic programming tool

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10692273B2 (en) * 2017-06-01 2020-06-23 Siemens Healthcare Gmbh In-context photorealistic 3D visualization for surgical decision support
CN108197632A (en) * 2017-12-04 2018-06-22 马艺菲 The identifying system and its method that distributed Freehandhand-drawing based on convolutional neural networks is drawn
CN107943474A (en) * 2017-12-05 2018-04-20 郑州大学西亚斯国际学院 A kind of programmed method of intelligent robot
CN108881446B (en) * 2018-06-22 2021-09-21 深源恒际科技有限公司 Artificial intelligence platform system based on deep learning
CN108635887A (en) * 2018-07-13 2018-10-12 王威立 A kind of hot-swappable building block system programing system

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104503754A (en) * 2014-12-16 2015-04-08 江南大学 Programming and compiling design method in robot graphical programming system
CN107169506A (en) * 2017-04-14 2017-09-15 微梦创科网络科技（中国）有限公司 Random assortment method and device based on assembled classifier
US20190042938A1 (en) * 2017-05-19 2019-02-07 Tata Consultancy Services Limited Simultaneous multi-class learning for data classification
CN207233274U (en) * 2017-07-06 2018-04-13 深圳市魔块智能有限公司 Wireless programmable electronic modular system
CN108460399A (en) * 2017-12-29 2018-08-28 华南师范大学 A kind of child building block builds householder method and system
CN108984693A (en) * 2018-07-04 2018-12-11 深圳点猫科技有限公司 A kind of sharing method and system of the program based on artificial intelligence
CN109284100A (en) * 2018-07-30 2019-01-29 深圳点猫科技有限公司 The rendering method and system of graphical programming language for subject cognitive tool
CN109344959A (en) * 2018-08-27 2019-02-15 联想(北京)有限公司 Neural network training method, nerve network system and computer system
CN109978000A (en) * 2019-02-19 2019-07-05 深圳点猫科技有限公司 A kind of classification AI implementation method and electronic equipment based on graphic programming tool

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113033297A (en) * 2021-02-08 2021-06-25 深圳市优必选科技股份有限公司 Object programming method, device, equipment and storage medium
CN113033297B (en) * 2021-02-08 2023-11-21 深圳市优必选科技股份有限公司 Method, device, equipment and storage medium for programming real object

Also Published As

Publication number Publication date
CN109978000A (en) 2019-07-05

Similar Documents

Publication Publication Date Title
WO2020168690A1 (en) Ai implementation method for classification based on graphical programming tool, and electronic device
ȚĂRANU Data mining in healthcare: decision making and precision.
WO2018223194A1 (en) Systems and methods of prediction of injury risk with a training regime
Liu et al. Event-driven continuous STDP learning with deep structure for visual pattern recognition
CN110232280A (en) A kind of software security flaw detection method based on tree construction convolutional neural networks
CN107480575A (en) The training method of model, across age face identification method and corresponding device
US20210406687A1 (en) Method for predicting attribute of target object based on machine learning and related device
CN107545151A (en) A kind of medicine method for relocating based on low-rank matrix filling
CN105825064A (en) Daily traditional Chinese medicine therapy intelligent consultancy system
CN107590473A (en) A kind of human face in-vivo detection method, medium and relevant apparatus
CN106529110A (en) Classification method and equipment of user data
CN109935337A (en) A kind of medical record lookup method and system based on similarity measurement
CN106779053A (en) The knowledge point of a kind of allowed for influencing factors and neutral net is known the real situation method
CN107944398A (en) Based on depth characteristic association list diagram image set face identification method, device and medium
CN111785366A (en) Method and device for determining patient treatment scheme and computer equipment
Raju et al. Optimized building of machine learning technique for thyroid monitoring and analysis
CN105809119A (en) Sparse low-rank structure based multi-task learning behavior identification method
Progga et al. A cnn based model for venomous and non-venomous snake classification
Wei et al. (Retracted) Image analysis and pattern recognition method of three-dimensional process in physical education teaching based on big data
CN108717548A (en) A kind of increased Activity recognition model update method of facing sensing device dynamic and system
CN104077524B (en) Training method and viruses indentification method and device for viruses indentification
Nikitin et al. Decision rule elicitation for domain adaptation
Mahmoodabadi et al. CADICA: Diagnosis of coronary artery disease using the imperialist competitive algorithm
CN108304856A (en) Image classification method based on cortex thalamus computation model
Alzahrani Bioinspired Image Processing Enabled Facial Emotion Recognition Using Equilibrium Optimizer With a Hybrid Deep Learning Model

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: 19916413

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 11.01.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 19916413

Country of ref document: EP

Kind code of ref document: A1