WO2022222020A1 - 一种用于流量分类的神经网络架构自动搜索方法及装置 - Google Patents

一种用于流量分类的神经网络架构自动搜索方法及装置 Download PDF

Info

Publication number
WO2022222020A1
WO2022222020A1 PCT/CN2021/088293 CN2021088293W WO2022222020A1 WO 2022222020 A1 WO2022222020 A1 WO 2022222020A1 CN 2021088293 W CN2021088293 W CN 2021088293W WO 2022222020 A1 WO2022222020 A1 WO 2022222020A1
Authority
WO
WIPO (PCT)
Prior art keywords
neural network
network architecture
traffic
search
searching
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2021/088293
Other languages
English (en)
French (fr)
Inventor
林鹏
叶可江
须成忠
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
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 Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Priority to PCT/CN2021/088293 priority Critical patent/WO2022222020A1/zh
Publication of WO2022222020A1 publication Critical patent/WO2022222020A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • G06N3/082Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/0464Convolutional networks [CNN, ConvNet]
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • G06N3/09Supervised learning
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • G06N3/0985Hyperparameter optimisation; Meta-learning; Learning-to-learn

Definitions

  • the present application relates to the field of information technology, and in particular, to an automatic search method and device for a neural network architecture for traffic classification.
  • Network traffic classification technology is an important research problem in the network field. It plays a very key role in intrusion detection, network anomaly detection and QoS assurance. As a very promising technology in recent years, deep neural network has achieved good results in the field of traffic classification. However, the current deep neural networks all require manual design and are difficult to adapt to different traffic classification tasks. These different traffic classification tasks may have different focuses, and the traffic patterns in different enterprises are also different. Designing a suitable neural network architecture is very expensive.
  • the linear layer and the Softmax layer are generally combined at the end for classification.
  • the present application provides an automatic search method and device for a neural network architecture for traffic classification, aiming at realizing automatic neural network architecture search.
  • an automatic search method for a neural network architecture for traffic classification comprising the following steps:
  • search for the optimal neural network structure including searching for the optimal Cell structure, repeating the search and stacking the same Cell structure, and forming the searched Cell structure into the entire neural network structure.
  • the method further includes:
  • the technical solutions adopted in the embodiments of the present application further include: in acquiring network traffic data samples, preprocessing the data samples and converting the data samples into a traffic data set in the form of a two-dimensional matrix, including:
  • Step 1 perform data extraction at network flow granularity
  • Step 2 carry out data extraction at packet granularity
  • Step 3 splicing the data extracted at the network flow granularity and the data extracted at the data packet granularity along the channel dimension to form a tensor
  • Step 4 Perform the operations from Step 1 to Step 3 above for all network flows to construct a formatted traffic data set.
  • the technical solution adopted in the embodiment of the present application further includes: dividing the traffic data set into training set, verification set and test set respectively includes:
  • the training set, validation set and test set contain the same proportion of traffic of each category; among them,
  • the training set is used to update the network weight parameters in this structure
  • the validation set is used to search for the best network structure parameters
  • test set is used to retrain using the training set plus the validation set after the search is complete, and to evaluate the final results on the test set.
  • the technical solutions adopted in the embodiments of the present application further include: the optional operation of constructing the search space includes: limiting the scope of the search neural network architecture space.
  • the technical solutions adopted in the embodiments of the present application further include: the optional operation of the search space of the neural network architecture is set as: 3x3 depthwise separable convolution, 5x5 depthwise separable convolution, 3x3 hole convolution, 5x5 hole convolution, 3x3 Maximum pooling and 3x3 average pooling.
  • searching for the optimal neural network architecture includes:
  • Step 1 Build a supernet; in which, the operation of extending eight edges in the supernet, and each edge has a different weight;
  • Step 2 Use gradient descent on the validation set loss to update the neural network architecture parameters
  • Step 3 Use the gradient descent method to update the operation weight parameters on the training set loss
  • Step 4 Repeat steps 2 and 3 until the neural network architecture training is completed
  • Step 5 After the training is completed, keep the edge with the largest node weight.
  • the technical solution adopted in the embodiment of the present application also includes: adding spatial attention and channel attention to the output position of each Cell, including:
  • Step 1 The channel attention weight is obtained by adding channel attention to the output of the searched Cell structure
  • Step 2 The spatial attention weight is obtained by adding spatial attention to the output of the searched Cell structure
  • Step 3 Perform matrix multiplication of the obtained channel attention weight and spatial attention weight to obtain a new output of attention weighting.
  • a neural network architecture automatic search device for traffic classification comprising:
  • Data conversion module used to obtain network traffic data samples, preprocess the data samples and convert them into a traffic data set in the form of a two-dimensional matrix;
  • Data division module used to divide the traffic data set into training set, validation set and test set respectively;
  • Optional operation building block optional operation used to build the search space of the neural network architecture after the traffic dataset is divided;
  • Search for the optimal network architecture module It is used to search for the optimal neural network architecture based on the optional operation of the search space; including, searching for the optimal Cell structure, repeating the search for stacking the same Cell structure, and forming the searched Cell structure into a whole Neural network architecture.
  • the device further includes:
  • Attention adding module used to add spatial attention and channel attention to the output position of each Cell
  • Retraining module used to save the searched Cell architecture parameters, and retrain the Cell architecture using a new training set that combines the training set and the validation set.
  • the neural network architecture automatic search method and device for traffic classification in the embodiments of the present application obtain network traffic data samples, preprocess the data samples, and transfer the data samples.
  • the traffic data set in the form of a two-dimensional matrix; the traffic data set is divided into training set, validation set and test set respectively; after the traffic data set is divided, the optional operation of constructing the search space of the neural network architecture; the search space-based optional operation
  • the selection operation is performed to search for the optimal neural network architecture; which includes, searching for the optimal Cell structure, repeatedly searching and stacking the same Cell structure, and forming the searched Cell structure into the entire neural network structure.
  • the present application can automatically search for the optimal neural network structure through a given candidate operation without manual participation.
  • FIG. 1 is a flowchart of an automatic search method for a neural network architecture for traffic classification according to an embodiment of the present application
  • Fig. 2 is the preferred flow chart of the neural network architecture automatic search method for traffic classification according to the embodiment of the present application
  • FIG. 3 is a block diagram of a neural network architecture automatic search device for traffic classification according to an embodiment of the present application
  • FIG. 4 is a preferred module diagram of a neural network architecture automatic search device for traffic classification according to an embodiment of the present application
  • FIG. 5 is a preferred module diagram of a neural network architecture automatic search device for traffic classification according to an embodiment of the present application
  • FIG. 6 is a schematic diagram of adding attention after the searched Cell structure described in an embodiment of the present application.
  • FIG. 7 is a diagram illustrating a calculation process of channel attention in an embodiment of the present application.
  • FIG. 8 is a diagram illustrating a calculation process of spatial attention according to an embodiment of the present application.
  • an automatic search method for a neural network architecture for traffic classification including the following steps:
  • S101 Obtain network traffic data samples, preprocess the data samples and convert them into a traffic data set in the form of a two-dimensional matrix;
  • S103 an optional operation of constructing a search space of the neural network architecture after the traffic data set is divided;
  • S104 Searching for an optimal neural network architecture based on an optional operation in the search space, which includes searching for an optimal Cell structure, repeatedly searching and stacking the same Cell structure, and forming the searched Cell structure into an entire neural network architecture.
  • the neural network architecture automatic search method and device for traffic classification in the embodiment of the present application obtains network traffic data samples, preprocesses the data samples and converts them into a traffic data set in the form of a two-dimensional matrix; and divides the traffic data sets into separate into a training set, a validation set and a test set; after the traffic data set is divided, an optional operation to construct the search space of the neural network architecture; an optional operation based on the search space to search for the optimal neural network architecture; including, searching for the most optimal neural network architecture. For an optimal Cell structure, repeat the search and stack the same Cell structure, and form the searched Cell structure into the entire neural network architecture.
  • the present application can automatically search for the optimal neural network structure through a given candidate operation without manual participation.
  • the method further includes:
  • S106 Save the searched Cell architecture parameters, and retrain the Cell architecture using a new training set formed by combining the training set and the validation set.
  • the spatial attention and channel attention mechanisms are combined into each cell structure searched, which enhances the feature extraction capability of the model.
  • spatial attention and channel attention are added to the output position of each cell structure searched to enhance the feature extraction ability of the model; the Cell structure obtained by adding spatial attention and channel attention training will be added. After the parameters are saved, continue to retrain until the training is complete.
  • the data samples are preprocessed and converted into a traffic data set in the form of a two-dimensional matrix, including:
  • Step 1 perform data extraction at network flow granularity
  • Step 2 carry out data extraction at packet granularity
  • Step 3 splicing the data extracted at the network flow granularity and the data extracted at the data packet granularity along the channel dimension to form a tensor
  • Step 4 Perform the operations from Step 1 to Step 3 above for all network flows to construct a formatted traffic data set.
  • Step 1 Perform data extraction at the network flow granularity, extract the previous byte of the network flow, and form The two-dimensional matrix m 1 of .
  • Step 2 perform data extraction at the data packet granularity, extract bytes from the first n data packets of a flow, extract the first ⁇ /n bytes of each data packet, total ⁇ bytes, and form The two-dimensional matrix m 2 of .
  • Step 3 splicing m 1 and m 2 along the channel dimension to form a 32*32*3 tensor.
  • Step 4 Perform the operations from Steps 1 to 3 above on all network flows to construct a formatted traffic data set.
  • dividing the traffic data set into training set, validation set and test set respectively includes:
  • the training set, validation set and test set contain the same proportion of traffic of each category; among them,
  • the training set is used to update the network weight parameters in this structure
  • the validation set is used to search for the best network structure parameters
  • test set is used to retrain using the training set plus the validation set after the search is complete, and to evaluate the final results on the test set.
  • the traffic dataset constructed above is divided, one part is used for training, one part is used for validation, and the remaining part is used for testing.
  • Each set contains the same proportion of each category of traffic.
  • the training set is used to update the network weight parameters in the structure, and the validation set is used to search for the best network structure parameters. After the search is completed, the training set and the validation set are used for retraining, and the final result is evaluated on the test set.
  • the optional operation of constructing the search space includes: limiting the scope of searching the neural network architecture space.
  • the optional operations for automatic neural architecture search are infinite, and to reduce complexity we need to limit the scope of the search space.
  • optional operations of the search space of the neural network architecture are set to: 3x3 depthwise separable convolution, 5x5 depthwise separable convolution, 3x3 atrous convolution, 5x5 atrous convolution, 3x3 max pooling and 3x3 averaging pooling.
  • Optional operations on the search space in this application include: 3x3 depthwise separable convolution, 5x5 depthwise separable convolution, 3x3 atrous convolution, 5x5 atrous convolution, 3x3 max pooling, 3x3 average pooling, identity and direct A total of eight operations.
  • searching for the optimal neural network architecture includes:
  • Step 1 Build a supernet; in which, the operation of extending eight edges in the supernet, and each edge has a different weight;
  • Step 2 Use gradient descent on the validation set loss to update the neural network architecture parameters
  • Step 3 Use the gradient descent method to update the operation weight parameters on the training set loss
  • Step 4 Repeat steps 2 and 3 until the neural network architecture training is completed
  • Step 5 After the training is completed, keep the edge with the largest node weight.
  • This search method belongs to microscopic search, that is, to search for an optimal Cell structure, and then repeatedly stack the same Cell to form the entire network.
  • Each Cell contains four nodes, and the nodes represent the feature map at a certain moment.
  • the edges between nodes represent some kind of optional operation, and the total number of optional operations is eight kinds.
  • a directed acyclic graph that is, the structure of a Cell, as shown in Figure 5.
  • Step 1 Build a supernet through the supernet construction formula, expand the operation of eight edges in the supernet, and each edge has a different weight;
  • the supernet construction formula is as follows:
  • o(x) is one of the operations in the optional operation set, that is, it represents one of the eight sides; Refers to the weight of the edge o(x), (i, j) defines that the edge belongs between the i and j nodes; Represents the weighted operation of eight edges between nodes i and i, and a feature map can be obtained through the calculation of the weighted operation; in addition, the formula uses the Softmax function to processed;
  • Step 2 Use the gradient descent method to update the neural network architecture parameter ⁇ on the validation set loss, the formula is as follows:
  • Step 3 Use the gradient descent method to update the operation weight parameter w on the training set loss, the formula is as follows:
  • Step 4 Repeat steps 2 and 3 until the neural network architecture training is completed
  • Step 5 After the training is completed, the maximum weight between the reserved nodes (i, j) is achieved through the node reservation formula The edge of ; the node retention formula is:
  • each node only retains connections to the previous two nodes with the largest weighted edges.
  • adding spatial attention and channel attention to the output position of each Cell includes:
  • Step 1 The channel attention weight is obtained by adding channel attention to the output of the searched Cell structure
  • Step 2 The spatial attention weight is obtained by adding spatial attention to the output of the searched Cell structure
  • Step 3 Perform matrix multiplication of the obtained channel attention weight and spatial attention weight to obtain a new output of attention weighting.
  • Step 1 Add channel attention to the output of the searched Cell structure, as shown in Figure 7; which includes, first, global average pooling and global maximum pooling are used for x with dimension h*w*C to obtain two A 1*1*C tensor, and then use a shared MLP layer to further extract features from the two tensors, and add them to obtain the channel attention weight M c through the Sigmoid function.
  • Step 2 Add spatial attention to the output of the searched Cell structure, as shown in Figure 8; which includes, first, global average pooling and global maximum pooling are adopted for x with dimension h*w*C to obtain two h*w*1 tensors, and then use a convolution kernel to convert the two tensors into a h*w*1 feature map, and then process the Sigmoid function to obtain the spatial attention weight M s .
  • Step 3 Perform matrix multiplication on the original output dimension x through the matrix multiplication formula with the obtained channel attention weight M c and spatial attention weight M s to obtain a new output x' of attention weighting.
  • the matrix multiplication formula is as follows:
  • a neural network architecture automatic search device for traffic classification is provided, referring to FIG. 3 to FIG. 8 , including:
  • Data conversion module used to obtain network traffic data samples, preprocess the data samples and convert them into a traffic data set in the form of a two-dimensional matrix;
  • Data division module used to divide the traffic data set into training set, validation set and test set respectively;
  • Optional operation building block optional operation used to build the search space of the neural network architecture after the traffic dataset is divided;
  • Search for the optimal network architecture module It is used to search for the optimal neural network architecture based on the optional operation of the search space; including, searching for the optimal Cell structure, repeating the search for stacking the same Cell structure, and forming the searched Cell structure into a whole Neural network architecture.
  • the neural network architecture automatic search method and device for traffic classification in the embodiment of the present application and the data conversion module: used to obtain network traffic data samples, preprocess the data samples, and convert them into a traffic data set in the form of a two-dimensional matrix;
  • Data division module used to divide the traffic data set into training set, validation set and test set respectively;
  • optional operation building module optional operation used to construct the search space of the neural network architecture after the traffic data set is divided;
  • search Optimal network architecture module It is used to search for the optimal neural network architecture based on the optional operation of the search space; including, searching for the optimal Cell structure, repeatedly searching and stacking the same Cell structure, and forming the searched Cell structure into the entire neural network.
  • Network Architecture The present application can automatically search for the optimal neural network structure through a given candidate operation without manual participation.
  • the device further includes:
  • Attention adding module used to add spatial attention and channel attention to the output position of each Cell
  • Retraining module used to save the searched Cell architecture parameters, and retrain the Cell architecture using a new training set that combines the training set and the validation set.
  • the spatial attention and channel attention mechanisms are combined into each cell structure searched, which enhances the feature extraction capability of the model.
  • spatial attention and channel attention are added to the output position of each cell structure searched to enhance the feature extraction ability of the model; the Cell structure obtained by adding spatial attention and channel attention training will be added. After the parameters are saved, continue to retrain until the training is complete.
  • the neural network architecture automatic search method and device for traffic classification in the embodiment of the present application obtains network traffic data samples, preprocesses the data samples and converts them into a traffic data set in the form of a two-dimensional matrix; It is divided into training set, validation set and test set respectively; after the division of the traffic data set, the optional operation of constructing the search space of the neural network architecture; the optional operation based on the search space, searching for the optimal neural network architecture; including, Search for the optimal Cell structure, repeat the search and stack the same Cell structure, and form the searched Cell structure into the entire neural network architecture.
  • the present application can automatically search for the optimal neural network structure through a given candidate operation without manual participation.
  • the application uses two-dimensional data matrix extraction with two granularities, thereby forming a dual-channel input.
  • This application adds spatial attention and channel attention to each Cell structure to enhance feature extraction capabilities.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

一种用于流量分类的神经网络架构自动搜索方法及装置。该方法及装置获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集(S101);将流量数据集分别划分成训练集、验证集和测试集(S102);在流量数据集划分后,构建神经网络架构的搜索空间的可选操作(S103);基于搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构(S104)。上述方法能够通过给定的候选操作自动地搜索出最优的神经网络结构,而无需人工的参与。

Description

一种用于流量分类的神经网络架构自动搜索方法及装置 技术领域
本申请涉及信息技术领域,具体而言,涉及一种用于流量分类的神经网络架构自动搜索方法及装置。
背景技术
网络流量分类技术是网络领域中一个重要的研究问题,它在入侵检测、网络异常检测和QoS保证中都起到非常关键的作用。深度神经网络作为近年来一种非常有前途的技术,已经在流量分类领域中取得了很好的效果。然而,目前的深度神经网络均需要人工设计,难以适应不同的流量分类任务,这些不同的流量分类任务可能存在不同的侧重点,且不同企业中的流量模式也各不相同,为每个任务单独设计一套适用的神经网络架构是非常高成本的。
传统的流量分类技术是基于端口号和深度包检测的,这种方法基于预先设定的规则来对待分类流量进行匹配,从而将其划分到已知的流量模式中。基于机器学习的方法使得流量分类技术能更加智能地去识别其内在的模式,这类方法一般先需要对网络流量进行特征化处理,将其转变为一组流量特征向量,然后结合常用的机器学习方法,如SVM、随机森林、KNN等进行分类。深度神经网络的出现使得端到端的流量分类成为可能,这类方法不强制要求事先将网络流量特征化,也可以将原始的网络数据进行简单编码后输送到神经网络中。神经网络先将这些编码映射到一组高维向量,然后对其进行卷积、池化、递归计算等操作,从而提取出隐藏特征。一般而言为了加强学习到的特征表达,会 加深网络的层数,这就是“深度神经网络”的由来。在多层的特征提取操作后,一般会在最后结合线性层和Softmax层来进行分类。
传统的流量分类技术基于端口号和深度包检测,这类方法很容易被端口欺骗、流量加密等技术所绕过。基于机器学习的方法需要先对网络流量进行特征化,然而这需要非常大量的专家先验知识,且人工设计出来的特征不一定能保证有代表性和全面性。深度学习方法虽然可以降低专家知识的门槛,然而设计有效的神经网络架构本身就是一件有挑战性的事情。另外,要找出哪一种神经网络架构最适合当前的任务,需要大量的实验和重设计,这需要耗费巨大的人力物力。
发明内容
本申请提供了一种用于流量分类的神经网络架构自动搜索方法及装置,目的在于实现自动的神经网络架构的搜索。
为了解决上述问题,本申请提供了如下技术方案:
一方面,提供一种用于流量分类的神经网络架构自动搜索方法,包括以下步骤:
获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集;
将流量数据集分别划分成训练集、验证集和测试集;
在流量数据集划分后,构建神经网络架构的搜索空间的可选操作;
基于搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神 经网络架构。
本申请实施例采取的技术方案还包括:方法还包括:
在每个Cell的输出位置加上空间注意力和通道注意力;
保存搜索出来的Cell架构参数,并使用训练集和验证集合并成的新训练集进行重新训练Cell架构。
本申请实施例采取的技术方案还包括:在获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集中包括:
步骤一:在网络流粒度进行数据提取;
步骤二:在数据包粒度进行数据提取;
步骤三:将在网络流粒度提取数据和在数据包粒度提取的数据沿通道维度进行拼接组成张量;
步骤四:对所有网络流进行上述第一步至第三步的操作,构建格式化的流量数据集。
本申请实施例采取的技术方案还包括:在将流量数据集分别划分成训练集、验证集和测试集中包括:
训练集、验证集及测试集中含有相同比例的各类别流量;其中,
训练集用于更新该结构中的网络权重参数;
验证集用于搜索最佳的网络结构参数;
测试集用于在搜索完成后使用训练集加验证集进行重新训练、并在测试集上评测最终结果。
本申请实施例采取的技术方案还包括:构建搜索空间的可选操作中包括:限定搜索神经网络架构空间的范围。
本申请实施例采取的技术方案还包括:神经网络架构的搜索空间的可选操 作为设置为:3x3深度可分离卷积、5x5深度可分离卷积、3x3空洞卷积、5x5空洞卷积、3x3最大化池化及3x3平均池化。
本申请实施例采取的技术方案还包括:在基于搜索空间的可选操作,搜索最优的神经网络架构中包括:
步骤一:构建超网;其中,在超网中扩展八条边的操作,且每条边有不同的权重;
步骤二:在验证集损失上使用梯度下降法更新神经网络架构参数;
步骤三:在训练集损失上使用梯度下降法更新操作权重参数;
步骤四:重复步骤二、步骤三直至神经网络架构训练完成;
步骤五:训练完成后,保留节点权重最大的那条边。
本申请实施例采取的技术方案还包括:在每个Cell的输出位置加上空间注意力和通道注意力中包括:
步骤一:通过对搜索出的Cell结构的输出后面加入通道注意力进行处理,得到通道注意力权重;
步骤二:通过对搜索出的Cell结构的输出后面加入空间注意力进行处理,得到空间注意力权重;
步骤三:将得到的通道注意力权重和空间注意力权重进行矩阵乘法,得到注意力加权的新输出。
另一方面,提供一种用于流量分类的神经网络架构自动搜索装置,装置包括:
数据转换模块:用于获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集;
数据分成模块:用于将流量数据集分别划分成训练集、验证集和测试集;
可选操作构建模块:用于在流量数据集划分后,构建神经网络架构的搜索空间的可选操作;
搜索最优网络架构模块:用于基于搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构。
本申请实施例采取的技术方案还包括:装置还包括:
注意力加入模块:用于在每个Cell的输出位置加上空间注意力和通道注意力;
重新训练模块:用于保存搜索出来的Cell架构参数,并使用训练集和验证集合并成的新训练集进行重新训练Cell架构。
相对于现有技术,本申请实施例产生的有益效果在于:本申请实施例中的用于流量分类的神经网络架构自动搜索方法及装置,获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集;将流量数据集分别划分成训练集、验证集和测试集;在流量数据集划分后,构建神经网络架构的搜索空间的可选操作;基于搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构。本申请能够通过给定的候选操作自动地搜索出最优的神经网络结构,而无需人工的参与。
附图说明
此处所说明的附图用来提供对本申请的进一步理解,构成本申请的一部分,本申请的示意性实施例及其说明用于解释本申请,并不构成对本申请的不当限定。在附图中:
图1是本申请实施例用于流量分类的神经网络架构自动搜索方法的流程图;
图2是本申请实施例用于流量分类的神经网络架构自动搜索方法的优选流程图;
图3是本申请实施例用于流量分类的神经网络架构自动搜索装置的模块图;
图4是本申请实施例用于流量分类的神经网络架构自动搜索装置的优选模块图;
图5是本申请实施例用于流量分类的神经网络架构自动搜索装置的优选模块图;
图6是本申请实施例描述在搜索到的Cell结构后面加入注意力的原理图;
图7是本申请实施例描述通道注意力的计算过程图;
图8是本申请实施例描述空间注意力的计算过程图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处所描述的具体实施例仅用以解释本申请,并不用于限定本申请。
需要说明的是,本申请的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象,而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换,以便这里描述的本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施。此外,术语“包 括”和“具有”以及他们的任何变形,意图在于覆盖不排他的包含,例如,包含了一系列步骤或单元的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或单元,而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它步骤或单元。
实施例1
参见图1至图8,根据本申请一实施例,提供了一种用于流量分类的神经网络架构自动搜索方法,包括以下步骤:
S101:获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集;
S102:将流量数据集分别划分成训练集、验证集和测试集;
S103:在流量数据集划分后,构建神经网络架构的搜索空间的可选操作;
S104:基于搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构。
本申请实施例中的用于流量分类的神经网络架构自动搜索方法及装置,获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集;将流量数据集分别划分成训练集、验证集和测试集;在流量数据集划分后,构建神经网络架构的搜索空间的可选操作;基于搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构。本申请能够通过给定的候选操作自动地搜索出最优的神经网络结构,而无需人工的参与。
实施例中,方法还包括:
S105:在每个Cell的输出位置加上空间注意力和通道注意力;
S106:保存搜索出来的Cell架构参数,并使用训练集和验证集合并成的新训练集进行重新训练Cell架构。
本申请技术中将空间注意力、通道注意力机制结合在搜索到的每个Cell结构中,增强了模型的特征提取能力。
本申请技术中在搜索到的每个Cell结构的输出位置中,加上空间注意力和通道注意力,以增强模型的特征提取能力;将加上空间注意力和通道注意力训练得到的Cell架构参数保存下来之后,继续重新训练直至训练完。
实施例中,在获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集中包括:
步骤一:在网络流粒度进行数据提取;
步骤二:在数据包粒度进行数据提取;
步骤三:将在网络流粒度提取数据和在数据包粒度提取的数据沿通道维度进行拼接组成张量;
步骤四:对所有网络流进行上述第一步至第三步的操作,构建格式化的流量数据集。
下面以具体实施例,对本申请的将数据样本进行预处理并转成二维矩阵形式的流量数据集进行详细说明:
步骤一,在网络流粒度进行数据提取,将网络流的前个字节提取出来,组成
Figure PCTCN2021088293-appb-000001
的二维矩阵m 1
步骤二,在数据包粒度进行数据提取,对一个流的前η个数据包进行字节提取,每个数据包提取前α/η个字节,总计α字节,组成
Figure PCTCN2021088293-appb-000002
的二维矩阵m 2
步骤三,将m 1和m 2沿通道维度进行拼接,组成32*32*3的张量。
步骤四,对所有网络流进行上述第一步至第三步的操作,构建格式化的流 量数据集。
实施例中,在将流量数据集分别划分成训练集、验证集和测试集中包括:
训练集、验证集及测试集中含有相同比例的各类别流量;其中,
训练集用于更新该结构中的网络权重参数;
验证集用于搜索最佳的网络结构参数;
测试集用于在搜索完成后使用训练集加验证集进行重新训练、并在测试集上评测最终结果。
将上述构建的流量数据集进行划分,一部分用于训练,一部分用于验证,剩余的部分用于测试。每个集中含有相同比例的各类别流量。训练集用于更新该结构中的网络权重参数,验证集用于搜索最佳的网络结构参数,在搜索完成后使用训练集结合验证集进行重新训练,并在测试集上评测最终结果。
实施例中,构建搜索空间的可选操作中包括:限定搜索神经网络架构空间的范围。自动神经架构搜索的可选操作是无限的,为了降低复杂度我们需要限定搜索空间的范围。
实施例中,神经网络架构的搜索空间的可选操作为设置为:3x3深度可分离卷积、5x5深度可分离卷积、3x3空洞卷积、5x5空洞卷积、3x3最大化池化及3x3平均池化。本申请中搜索空间的可选操作包括:3x3深度可分离卷积、5x5深度可分离卷积、3x3空洞卷积、5x5空洞卷积、3x3最大化池化、3x3平均池化、恒等及直连共八个操作。
实施例中,在基于搜索空间的可选操作,搜索最优的神经网络架构中包括:
步骤一:构建超网;其中,在超网中扩展八条边的操作,且每条边有不同的权重;
步骤二:在验证集损失上使用梯度下降法更新神经网络架构参数;
步骤三:在训练集损失上使用梯度下降法更新操作权重参数;
步骤四:重复步骤二、步骤三直至神经网络架构训练完成;
步骤五:训练完成后,保留节点权重最大的那条边。
本搜索方法属于微观搜索,即搜索一个最优的Cell结构,然后重复堆叠相同的Cell形成整个网络。每一个Cell中包含四个节点,节点代表某一个时刻的特征图。节点与节点之间的边代表某种可选操作,总的可选操作数有种,本例中为八种。最终需要搜索出一个有向无环图,即一个Cell的结构,如图5所示。
下面以具体实施例,对本申请搜索最优的神经网络架构的过程进行详细说明:
步骤一:通过超网构建公式构建超网,在超网中扩展八条边的操作,且每条边有不同的权重;超网构建公式如下:
Figure PCTCN2021088293-appb-000003
其中,o(x)是可选操作集合中的其中一个操作,即代表八条边中的某一条边;
Figure PCTCN2021088293-appb-000004
指的是该边o(x)的权重,(i,j)限定了该边是属于i和j节点之间的;
Figure PCTCN2021088293-appb-000005
表示节点i和i之间的八条边加权操作,通过该加权操作的计算能够得到一个特征图;另外,该公式使用了Softmax函数对
Figure PCTCN2021088293-appb-000006
进行了处理;
步骤二:在验证集损失上使用梯度下降法更新神经网络架构参数α,公式如下:
Figure PCTCN2021088293-appb-000007
步骤三:在训练集损失上使用梯度下降法更新操作权重参数w,公式如下:
Figure PCTCN2021088293-appb-000008
步骤四:重复步骤二、步骤三直至神经网络架构训练完成;
步骤五:训练完成后,通过节点保留公式实现保留节点(i,j)之间权重最大
Figure PCTCN2021088293-appb-000009
的那条边;节点保留公式为:
Figure PCTCN2021088293-appb-000010
其中,每个节点只保留与前面两个具有最大权重边的节点的连接。
实施例中,在每个Cell的输出位置加上空间注意力和通道注意力中包括:
步骤一:通过对搜索出的Cell结构的输出后面加入通道注意力进行处理,得到通道注意力权重;
步骤二:通过对搜索出的Cell结构的输出后面加入空间注意力进行处理,得到空间注意力权重;
步骤三:将得到的通道注意力权重和空间注意力权重进行矩阵乘法,得到注意力加权的新输出。
下面以具体实施例,如图6和图8所示,对本申请每个Cell的输出位置加上空间注意力和通道注意力的过程进行详细说明:
步骤一:对搜索出的Cell结构的输出后面加入通道注意力,如图7所示;其中包括,首先对维度为h*w*C的x分别采取全局平均池化以及全局最大池化得到两个1*1*C的张量,然后使用一个共享的MLP层对该两个张量进一步提取特征,并将它们相加后经过Sigmoid函数得到通道注意力权重M c
步骤二:对搜索出的Cell结构的输出后面加入空间注意力,如图8所示;其中包括,首先对维度为h*w*C的x分别采取全局平均池化以及全局最大池化得到两个h*w*1的张量,然后使用一个卷积核将该两个张量转换为一个h*w*1的特征图,再经过Sigmoid函数处理后得到空间注意力权重M s
步骤三:将得到的通道注意力权重M c和空间注意力权重M s通过矩阵乘法公式在原输出维度x上进行矩阵乘法,得到注意力加权的新输出x',矩阵乘法公式如下所示:
x'=M s×(M c×x)
实施例2
根据本申请的另一实施例,提供了一种用于流量分类的神经网络架构自动搜索装置,参见图3至图8,包括:
数据转换模块:用于获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集;
数据分成模块:用于将流量数据集分别划分成训练集、验证集和测试集;
可选操作构建模块:用于在流量数据集划分后,构建神经网络架构的搜索空间的可选操作;
搜索最优网络架构模块:用于基于搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构。
本申请实施例中的用于流量分类的神经网络架构自动搜索方法及装置,数据转换模块:用于获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集;数据分成模块:用于将流量数据集分别划分成训练集、验证集和测试集;可选操作构建模块:用于在流量数据集划分后,构建神经网络架构的搜索空间的可选操作;搜索最优网络架构模块:用于基于搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构。本申请能够通过给定的候选操作自动地搜索出最优的神经网络结构,而无需人工的参 与。
实施例中,装置还包括:
注意力加入模块:用于在每个Cell的输出位置加上空间注意力和通道注意力;
重新训练模块:用于保存搜索出来的Cell架构参数,并使用训练集和验证集合并成的新训练集进行重新训练Cell架构。
本申请技术中将空间注意力、通道注意力机制结合在搜索到的每个Cell结构中,增强了模型的特征提取能力。
本申请技术中在搜索到的每个Cell结构的输出位置中,加上空间注意力和通道注意力,以增强模型的特征提取能力;将加上空间注意力和通道注意力训练得到的Cell架构参数保存下来之后,继续重新训练直至训练完。
本申请的有益效果在于:
1.本申请实施例中的用于流量分类的神经网络架构自动搜索方法及装置,获取网络流量数据样本,将数据样本进行预处理并转成二维矩阵形式的流量数据集;将流量数据集分别划分成训练集、验证集和测试集;在流量数据集划分后,构建神经网络架构的搜索空间的可选操作;基于搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构。本申请能够通过给定的候选操作自动地搜索出最优的神经网络结构,而无需人工的参与。
2.本申请使用的是两个粒度的数据二维矩阵提取,从而形成双通道的输入。
3.本申请在每个Cell结构中加入了空间注意力和通道注意力以增强特征提取能力。
对所公开的实施例的上述说明,使本领域专业技术人员能够实现或使用本申请。对这些实施例的多种修改对本领域的专业技术人员来说将是显而易见的,本文中所定义的一般原理可以在不脱离本申请的精神或范围的情况下,在其它实施例中实现。因此,本申请将不会被限制于本文所示的这些实施例,而是要符合与本文所公开的原理和新颖特点相一致的最宽的范围。

Claims (10)

  1. 一种用于流量分类的神经网络架构自动搜索方法,其特征在于,包括以下步骤:
    获取网络流量数据样本,将所述数据样本进行预处理并转成二维矩阵形式的流量数据集;
    将所述流量数据集分别划分成训练集、验证集和测试集;
    在所述流量数据集划分后,构建神经网络架构的搜索空间的可选操作;
    基于所述搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构。
  2. 根据权利要求1所述的用于流量分类的神经网络架构自动搜索方法,其特征在于,所述方法还包括:
    在每个Cell的输出位置加上空间注意力和通道注意力;
    保存搜索出来的Cell架构参数,并使用所述训练集和所述验证集合并成的新训练集进行重新训练Cell架构。
  3. 根据权利要求2所述的用于流量分类的神经网络架构自动搜索方法,其特征在于,在获取网络流量数据样本,将所述数据样本进行预处理并转成二维矩阵形式的流量数据集中包括:
    步骤一:在网络流粒度进行数据提取;
    步骤二:在数据包粒度进行数据提取;
    步骤三:将在网络流粒度提取数据和在数据包粒度提取的数据沿通道维度进行拼接组成张量;
    步骤四:对所有网络流进行上述第一步至第三步的操作,构建格式化的流量数据集。
  4. 根据权利要求2所述的用于流量分类的神经网络架构自动搜索方法,其特征在于,在将所述流量数据集分别划分成训练集、验证集和测试集中包括:
    所述训练集、验证集及测试集中含有相同比例的各类别流量;其中,
    所述训练集用于更新该结构中的网络权重参数;
    所述验证集用于搜索最佳的网络结构参数;
    所述测试集用于在搜索完成后使用训练集加验证集进行重新训练、并在测试集上评测最终结果。
  5. 根据权利要求2所述的用于流量分类的神经网络架构自动搜索方法,其特征在于,所述构建搜索空间的可选操作中包括:限定搜索神经网络架构空间的范围。
  6. 根据权利要求5所述的用于流量分类的神经网络架构自动搜索方法,其特征在于,所述神经网络架构的搜索空间的可选操作为设置为:3x3深度可分离卷积、5x5深度可分离卷积、3x3空洞卷积、5x5空洞卷积、3x3最大化池化及3x3平均池化。
  7. 根据权利要求2所述的用于流量分类的神经网络架构自动搜索方法,其特征在于,所述在基于所述搜索空间的可选操作,搜索最优的神经网络架构中包括:
    步骤一:构建超网;其中,在所述超网中扩展八条边的操作,且每条边有不同的权重;
    步骤二:在所述验证集损失上使用梯度下降法更新神经网络架构参数;
    步骤三:在所述训练集损失上使用梯度下降法更新操作权重参数;
    步骤四:重复步骤二、步骤三直至所述神经网络架构训练完成;
    步骤五:训练完成后,保留节点权重最大的那条边。
  8. 根据权利要求6所述的用于流量分类的神经网络架构自动搜索方法,其特征在于,所述在每个Cell的输出位置加上空间注意力和通道注意力中包括:
    步骤一:通过对搜索出的所述Cell结构的输出后面加入通道注意力进行处理,得到通道注意力权重;
    步骤二:通过对搜索出的所述Cell结构的输出后面加入空间注意力进行处理,得到空间注意力权重;
    步骤三:将得到的通道注意力权重和空间注意力权重进行矩阵乘法,得到注意力加权的新输出。
  9. 一种用于流量分类的神经网络架构自动搜索装置,其特征在于,所述装置包括:
    数据转换模块:用于获取网络流量数据样本,将所述数据样本进行预处理并转成二维矩阵形式的流量数据集;
    数据分成模块:用于将所述流量数据集分别划分成训练集、验证集和测试集;
    可选操作构建模块:用于在所述流量数据集划分后,构建神经网络架构的搜索空间的可选操作;
    搜索最优网络架构模块:用于基于所述搜索空间的可选操作,搜索最优的神经网络架构;其中包括,搜索最优的Cell结构,重复搜索堆叠相同的Cell结构,将搜索的Cell结构形成整个神经网络架构。
  10. 根据权利要求9所述的用于流量分类的神经网络架构自动搜索装置,其特征在于,所述装置还包括:
    注意力加入模块:用于在每个Cell的输出位置加上空间注意力和通道注意力;
    重新训练模块:用于保存搜索出来的Cell架构参数,并使用所述训练集和所述验证集合并成的新训练集进行重新训练Cell架构。
PCT/CN2021/088293 2021-04-20 2021-04-20 一种用于流量分类的神经网络架构自动搜索方法及装置 Ceased WO2022222020A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/088293 WO2022222020A1 (zh) 2021-04-20 2021-04-20 一种用于流量分类的神经网络架构自动搜索方法及装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/088293 WO2022222020A1 (zh) 2021-04-20 2021-04-20 一种用于流量分类的神经网络架构自动搜索方法及装置

Publications (1)

Publication Number Publication Date
WO2022222020A1 true WO2022222020A1 (zh) 2022-10-27

Family

ID=83723631

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/088293 Ceased WO2022222020A1 (zh) 2021-04-20 2021-04-20 一种用于流量分类的神经网络架构自动搜索方法及装置

Country Status (1)

Country Link
WO (1) WO2022222020A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115760777A (zh) * 2022-11-21 2023-03-07 脉得智能科技(无锡)有限公司 基于神经网络结构搜索的桥本氏甲状腺炎诊断系统
CN116232694A (zh) * 2023-01-31 2023-06-06 清华大学深圳国际研究生院 轻量级网络入侵检测方法、装置、电子设备及存储介质
CN116423924A (zh) * 2023-03-31 2023-07-14 广东佛斯伯智能设备有限公司 纸板线湿部生产速度模型构建方法及生产速度预测方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111818052A (zh) * 2020-07-09 2020-10-23 国网山西省电力公司信息通信分公司 基于cnn-lstm的工控协议同源攻击检测方法
CN112053351A (zh) * 2020-09-08 2020-12-08 哈尔滨工业大学(威海) 基于神经网络架构搜索和注意力机制的肺部结节良恶性判别方法
US20200401899A1 (en) * 2019-06-20 2020-12-24 Google Llc Computationally efficient neural network architecture search
CN112215269A (zh) * 2020-09-27 2021-01-12 苏州浪潮智能科技有限公司 一种面向目标检测的模型构建方法、设备和神经网络架构
CN112651406A (zh) * 2020-12-18 2021-04-13 浙江大学 一种深度感知和多模态自动融合的rgb-d显著性目标检测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200401899A1 (en) * 2019-06-20 2020-12-24 Google Llc Computationally efficient neural network architecture search
CN111818052A (zh) * 2020-07-09 2020-10-23 国网山西省电力公司信息通信分公司 基于cnn-lstm的工控协议同源攻击检测方法
CN112053351A (zh) * 2020-09-08 2020-12-08 哈尔滨工业大学(威海) 基于神经网络架构搜索和注意力机制的肺部结节良恶性判别方法
CN112215269A (zh) * 2020-09-27 2021-01-12 苏州浪潮智能科技有限公司 一种面向目标检测的模型构建方法、设备和神经网络架构
CN112651406A (zh) * 2020-12-18 2021-04-13 浙江大学 一种深度感知和多模态自动融合的rgb-d显著性目标检测方法

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115760777A (zh) * 2022-11-21 2023-03-07 脉得智能科技(无锡)有限公司 基于神经网络结构搜索的桥本氏甲状腺炎诊断系统
CN115760777B (zh) * 2022-11-21 2024-04-30 脉得智能科技(无锡)有限公司 基于神经网络结构搜索的桥本氏甲状腺炎诊断系统
CN116232694A (zh) * 2023-01-31 2023-06-06 清华大学深圳国际研究生院 轻量级网络入侵检测方法、装置、电子设备及存储介质
CN116423924A (zh) * 2023-03-31 2023-07-14 广东佛斯伯智能设备有限公司 纸板线湿部生产速度模型构建方法及生产速度预测方法

Similar Documents

Publication Publication Date Title
CN112308158B (zh) 一种基于部分特征对齐的多源领域自适应模型及方法
CN108764063B (zh) 一种基于特征金字塔的遥感影像时敏目标识别系统及方法
CN111783831A (zh) 基于多源多标签共享子空间学习的复杂图像精确分类方法
WO2019001070A1 (zh) 一种基于邻接矩阵的连接信息规整系统、图特征提取系统、图分类系统和方法
WO2022222020A1 (zh) 一种用于流量分类的神经网络架构自动搜索方法及装置
CN115439679A (zh) 一种联合多注意及Transformer的高光谱图像分类方法
CN110993037A (zh) 一种基于多视图分类模型的蛋白质活性预测装置
CN112528873B (zh) 基于多级语义表征和语义计算的信号语义识别方法
CN113128308B (zh) 一种港口场景下的行人检测方法、装置、设备及介质
CN114463651A (zh) 基于超轻量级高效卷积神经网络的农作物病虫害识别方法
CN110210335A (zh) 一种行人重识别学习模型的训练方法、系统和装置
CN111339818A (zh) 一种人脸多属性识别系统
CN104463194A (zh) 一种人车分类方法及装置
CN111860584A (zh) 图分类方法和装置
CN114463340A (zh) 一种边缘信息引导的敏捷型遥感图像语义分割方法
CN107169566A (zh) 动态神经网络模型训练方法和装置
CN110414338B (zh) 基于稀疏注意力网络的行人再辨识方法
CN116884074A (zh) 一种基于混合注意力机制的轻量级人脸识别方法
CN114298854B (zh) 一种联合学习表示和对齐的弱监督用户身份链接方法
Nasra et al. Automated Malaria Detection Using Deep Learning and the Lacuna Dataset
CN114884896B (zh) 一种基于特征扩展和自动机器学习的移动应用流量感知方法
CN104598898A (zh) 一种基于多任务拓扑学习的航拍图像快速识别系统及其快速识别方法
CN115410223A (zh) 一种基于不变性特征提取的域泛化行人重识别方法
CN115272278A (zh) 一种用于遥感图像变化检测的变化检测模型的构建方法
CN115310500A (zh) 一种用于流量分类的神经网络架构自动搜索方法及装置

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: 21937271

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21937271

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 21937271

Country of ref document: EP

Kind code of ref document: A1

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 27.05.2024)

122 Ep: pct application non-entry in european phase

Ref document number: 21937271

Country of ref document: EP

Kind code of ref document: A1