149 results about "Pooling" patented technology

The invention provides a multi-receptive-field network abnormal flow detection method and system based on endogenous security attributes, and relates to the technical field of network security and abnormal flow intelligent detection. The method comprises the following steps: firstly, performing multi-scale flow representation, preprocessing and data enhancement on network flow data to obtain enhanced input flow data; local features are extracted through basic convolution, and local and global fusion features are obtained based on a double-branch network comprising a multi-receptive field convolution branch and a Mama-self-attention branch; deep fusion representation is formed through multi-round feature extraction and tensor fusion, and finally binary classification and fine-grained classification results are output through global pooling and a linear classification layer. According to the method, high-precision, high-robustness and high-real-time detection of the abnormal traffic of the complex network is realized under low calculation overhead.

The invention discloses a dynamic space-time diagram flow prediction method and system based on course learning, and relates to the technical field of supply chain logistics data analysis, and the method comprises the steps: building a space-time matrix based on historical multi-source data, generating a dynamic adjacent matrix through learning, and carrying out the smooth fusion through combining a static diagram, and forming a dynamic diagram structure. And then space and time features are respectively extracted by using a graph convolutional network and a gating loop unit, and deep interaction and fusion are realized through a bidirectional cross attention mechanism. A multi-dimensional difficulty estimator is innovatively introduced, the prediction difficulty of each training sample is quantified from three dimensions of space, time and time-space coupling, the selection sequence of the training samples is dynamically adjusted based on an adaptive course scheduler, and progressive learning is realized. And finally, feature representation is obtained through global pooling, and multi-step traffic prediction is realized by adopting a parallel independent decoder, so that error accumulation is avoided. According to the invention, prediction precision and model training efficiency in a complex supply chain logistics scene are effectively improved.

The invention discloses a Tor network exit traffic identification system and method fusing a multi-scale LSTM and a Transform network, and belongs to the technical field of anonymous network traffic analysis and network security. The system comprises five core components, namely a multi-scale feature extraction module, a feature fusion module, a global dependency modeling module, a dynamic weighted aggregation module and a classification module. The multi-scale feature extraction module adopts parallel bidirectional LSTM branches with different time resolutions to capture a microcosmic burst mode and a macroscopic session behavior at the same time; the feature fusion module unifies the scale features to the same time sequence length and splices the scale features; the global dependence modeling module utilizes a multi-head self-attention mechanism to learn long-distance time sequence dependence; the dynamic weighted aggregation module highlights a key time slice through adaptive weight pooling; and the classification module outputs website category labels. According to the system, the recognition accuracy on a GTT23 data set is remarkably improved compared with that of an existing method, and good recognition capability and robustness are shown for various flow defense mechanisms.

The invention discloses an NVMe storage array virtualization device based on an FPGA (Field Programmable Gate Array), which is characterized in that single I / O (Input / Output) virtualization (SR-IOV) is realized in the FPGA, and a plurality of NVMe virtual functions which can be directly accessed by a virtual machine are externally provided; mapping from a virtual volume logic address to a physical storage address is completed in hardware, striping management and RAID verification calculation are carried out on a plurality of NVMe solid state disks at the rear end, and storage resource pooling is achieved; the partial reconfigurable characteristic of the FPGA is utilized, I / O scheduling, data compression and prefetching strategies are dynamically adjusted, and the adaptability and performance efficiency of the system are improved. Hardware unloading of the whole I / O path is completed in the FPGA, intervention of a host CPU is not needed, delay is remarkably reduced, the throughput capacity is improved, service quality isolation and elastic expansion in a multi-tenant environment are supported, and an effective technical path is provided for constructing a next-generation high-performance and low-delay storage infrastructure.

The invention discloses a reverse knowledge distillation-based cross-stage feature fusion method for industrial anomaly detection and positioning, and relates to the technical field of knowledge distillation and transfer learning. The method comprises the following steps: step 1, constructing a teacher network encoder, a student network encoder and a decoder; 2, extracting input image features through a student network encoder, and reconstructing the features through a student decoder; step 3, student network cross-stage feature fusion design; 4, locally sensing a dynamic attention module LDA, and adding a sliding window and convolution to replace original global pooling; and step 5, a self-adaptive multi-scale feature fusion module CAAMS-FF establishes a dependency relationship of a context so as to realize a fine-grained high-quality feature reconstruction effect. And step 6, inputting an anomaly graph obtained by the teacher-student network into the segmentation sub-network to obtain a final anomaly detection and positioning result. While the calculation efficiency is maintained, the spatial positioning precision and the multi-scale adaptability are significantly improved.

The invention discloses a switching communication device, a switching communication method and a server, relates to the technical field of computers, and realizes equipment-level decoupling and resource pooling basis by actively managing a GPU (Graphics Processing Unit) through a second interface configured to be in a root complex mode. The on-chip management system and the switching logic module form a control core, a data switching strategy is dynamically executed, and conversion from passive forwarding to autonomous management is achieved. And the virtual port module performs software definition and dynamic allocation on network resources, so that the flexibility of resource utilization is improved. Meanwhile, MAC ports in two modes of direct connection and exchange are supported, so that the device can flexibly construct a high-performance direct connection network or a large-scale exchange network, and the limitation of fixed topology is broken through. And the data processing module ensures high efficiency of data conversion between different interface protocols. According to the device, the communication efficiency and schedulability between the GPUs are improved by integrating high-speed exchange, intelligent management, resource virtualization and flexible networking capabilities.

The invention discloses a main and distribution network collaborative analysis method, system and device and a storage medium, and belongs to the technical field of power system analysis and calculation. The method comprises the following steps: abstracting a main power distribution network integrated model into a heterogeneous graph and generating a fusion node feature matrix; a time-space diagram neural network model is constructed and trained, the model ensures that a prediction result conforms to a power system physical law by introducing physical information neural network constraints, and multi-task parallel prediction is completed by adopting a diagram self-attention pooling and multi-task learning architecture; performing cross-domain interaction verification on a collaborative analysis result output by the model, wherein the cross-domain interaction verification comprises boundary consistency verification and energy / power balance verification; and dynamically adjusting calculation granularity or model parameters according to a verification result, and accelerating a calculation process based on a linear self-attention mechanism. According to the method, the problems of model splitting and inconsistent calculation results caused by independent analysis of the traditional main and distribution networks are solved, and global consistent, physically credible, efficient and collaborative power grid analysis is realized.

The invention discloses an open source environment software hidden vulnerability patch identification method, which is based on a multi-stage architecture and collaborative relationship modeling, and provides a new patch group identification scheme: firstly, obtaining candidate code submission from an open source warehouse, extracting correlation characteristics of vulnerabilities and candidate code submission, including rule-based characteristics and semantic characteristics, calculating a correlation score and screening high-correlation submission; pairwise pairing the high-correlation submissions, and fusing multi-dimensional features to predict a cooperative relationship between the submissions; and finally, constructing an undirected graph based on the correlation score, dividing a maximum connected sub-graph, fusing the features in the group through maximum pooling, calculating the correlation with the vulnerability, and outputting an optimal patch group. Meanwhile, an existing patch identification technology based on sorting learning is combined, an enhancement method based on a submission cooperative relation is provided, ranking logic submitted by candidate codes is updated through internal association between code submission and by means of correlation between group vectors and vulnerabilities, and the identification precision in a multi-patch scene is improved.

The invention relates to the technical field of computer vision, in particular to a small target fine segmentation method fusing improved pyramid pooling and edge branch supervision, and the method comprises the steps: selecting a data set, carrying out the preprocessing of a data set image, carrying out the feature extraction of the preprocessed image through a backbone network MobilenetV3, and generating four sets of feature maps F1 to F4, the method comprises the steps of inputting F1 and F2 into an ABG module, generating feature maps F5 and F4 containing edge information, sending the feature maps F5 and F4 into an improved ASPP structure, obtaining a feature map F6 containing multi-scale context information, inputting F3 and F4 into an FCE module for feature fusion so as to supplement spatial details, finally splicing and fusing F5 and F6, enhancing through the FCE module, and gradually performing up-sampling back to the original size in combination with the supervision of the feature maps F3 and F4, so as to obtain the multi-scale contextual information. And a final segmented image is obtained. According to the invention, through integration of the segmentation network, full-link optimization of'feature extraction-edge enhancement-multi-scale modeling-feature calibration-precise decoding 'is realized.

The application discloses an edge computing processing system based on a regional comprehensive energy station, and the edge computing processing system comprises a data input module, a data processing module, a storage module, a communication module and a power conversion module; the data input module converts analog flow from the regional comprehensive energy station into data information; the data processing module performs task distribution on the data information in a time sequence pipeline mode; the storage module stores and calls the task distribution; the communication module transmits the task distribution to a server; and the power conversion module is used for supplying power to the edge computing processing system; wherein the neural network operation module comprises an input buffer unit, an intermediate calculation buffer unit, an average pooling unit, a control unit and a convolution operation unit; and the application solves the technical problems of complex hardware structure of an edge computing terminal and slow data processing task in the prior art.

The invention relates to the technical field of part identification, and discloses a part identification method and device based on an industrial large model, and the method comprises the steps: carrying out the image block division and embedded transformation of an original image of an industrial part, obtaining a first Token sequence, carrying out the two-dimensional discrete cosine transformation of the first Token sequence, and obtaining a second Token sequence; fusing the second Token sequence with a plurality of prototype feature vectors selected from a category prototype memory library to obtain a memory guide vector; performing gating modulation and residual connection processing on the second Token sequence based on the memory guide vector to obtain a third Token sequence; and performing iterative refinement on the third Token sequence to obtain a fourth Token sequence, and performing global pooling and classification prediction on the fourth Token sequence to obtain a category identification result of the industrial parts, thereby enhancing the discrimination of the part characteristics, and effectively solving the technical problem of difficult identification of small samples of rare parts.

The invention relates to the technical field of storage virtualization, in particular to a hospital business-oriented server and storage virtualization system integration and disaster recovery reconstruction method, which comprises the following steps of: simulating layered integrated construction deployment of a system, and constructing a resource pool infrastructure which is deeply coupled with calculation and storage and is optimized aiming at medical business load; on the basis of a progressive online migration technology of a business portrait, risk-controllable and smooth seamless migration of a business system from a physical platform to a virtual platform is realized; the old machine room is reconstructed into a degraded disaster recovery center, and the old machine room is transformed into a disaster recovery site which is economical and practical and meets a core business recovery target; the method has the beneficial effects that the resource efficiency and economy are remarkably improved, the average resource utilization rate is improved from less than 20% to more than 60% through deep resource pooling, the purchase quantity of new equipment is reduced, meanwhile, old assets are fully utilized, and the TCO is reduced.

The invention provides a Chinese intention recognition method based on a cue word template, and relates to the field of network security in computer science and technology. The method comprises the following steps: segmenting a to-be-recognized text to obtain a token sequence; replacing attitude word tokens in the token sequence with special tokens [MASK] through a cue word template, and respectively adding special tokens [CLS] and [SEP] at the left end and the right end to obtain an input token sequence; inputting the input token sequence into a multi-layer Transform encoder to obtain a text token vector, a [MASK] special token vector and a [CLS] special token vector; averagely pooling the text token vector and the [MASK] special token vector to obtain a text pooling vector and a [MASK] pooling vector; sequentially splicing the [CLS] special token vector, the text pooling vector and the [MASK] pooling vector to obtain a final representation vector of the to-be-recognized text; and inputting the final representation vector into the MLP, and identifying the Chinese intention of the to-be-identified text. According to the method, accurate recognition of the Chinese text intention in the social platform is realized.

This invention provides a power load forecasting method based on time series core fusion, belonging to the field of artificial intelligence technology. The method involves normalizing adjustable load data to obtain adjustable load data feature vectors; constructing an adjustable power load forecasting model using a multi-layer perception layer, a random pooling layer, and a fusion layer; inputting the adjustable load data feature vectors into the adjustable power load forecasting model; training the adjustable power load forecasting model based on an accuracy index; and finally, obtaining the adjustable power load forecasting result by performing inverse normalization calculation on the feature sequences containing the forecast group, thus completing the forecasting of adjustable power load. This invention solves the problem of low accuracy in adjustable power load forecasting results.

The application relates to a rotating machine residual life prediction method and system, and belongs to the field of mechanical life prediction. The method comprises the following steps: collecting signals capable of reflecting mechanical life of a rotating machine to be measured at least at two positions, extracting one-dimensional time characteristic data from the signals capable of reflecting mechanical life by using principal component analysis pooling, training a related vector machine by using the one-dimensional time characteristic data, so that the full life cycle of the rotating machine is divided into a period without obvious failure trend and a failure tendency period; training a deep separable convolution gate recurrent unit network by using data of the failure tendency period, and predicting the residual life of the rotating machine in the failure tendency period. The application can improve the prediction accuracy when the residual service life of the rotating machine is predicted.

The invention discloses an alarm event early warning method and device suitable for multiple frames and multiple platforms, equipment and a medium, and relates to the field of alarm event early warning, and the method comprises the steps: obtaining sample data in each target frame and each target platform, and carrying out the alignment and slicing of the sample data, and obtaining the processed data; determining a feature set of the processed data through a convolutional neural network of a preset model, determining an initial feature sequence, corresponding to the first target time point, in the feature set in a convolutional block attention module of the preset model, and performing average pooling operation and maximum pooling operation on the initial feature sequence to obtain a second target time point; determining a second target time point by using the obtained first target feature sequence and the second target feature sequence; and determining a target feature corresponding to the second target time point in the feature set, and constructing a corresponding feature baseline graph to predict a preset alarm event by using the feature baseline graph. Therefore, the accuracy of performing early warning on the alarm event in a multi-framework and multi-platform environment can be improved.

This invention discloses a method and apparatus for accelerating distributed training of large AI models based on model prediction. The method includes: constructing a hierarchical parameter dataset; training a prediction model for predicting model parameters of a target model using an evaluation method based on prediction results; computing nodes updating the model parameters of some layers of the target model and then pushing them to a parameter server; on the parameter server, aggregating the received model parameters and predicting the model parameters of the remaining layers of the target model using the prediction model; concatenating the predicted and aggregated model parameters into a complete model parameter set and pushing it to all computing nodes; and replacing the local model parameters with the received complete model parameter set. The prediction model adopts a "convolution + channel attention mechanism + pooling" structure. This invention can significantly reduce communication volume without introducing a large amount of computational overhead while maintaining the accuracy of the target model.

This invention discloses a method, apparatus, device, and storage medium for detecting SQL injection attacks. The method includes the following steps: based on an abstract syntax tree generated by parsing a received SQL statement, semantically and syntactically encoding the nodes in the abstract syntax tree to obtain semantic feature vectors and syntactic feature vectors for each node; concatenating and fusing the semantic feature vectors and syntactic feature vectors to generate initial features for each node; based on a constructed sparse heterogeneous graph, using a graph convolutional network to propagate the initial features between each node and its neighboring nodes to obtain updated features for each node; performing global average pooling on the updated features of the nodes to obtain a global representation vector of the SQL statement, and inputting the global representation vector into a classifier to output the SQL injection detection result. This application can significantly reduce the computational resource requirements of the model while maintaining high detection accuracy.

The present disclosure is a lightweight target detection method based on feature enhancement and multi-branch fusion, comprising: feature extraction and enhancement: extracting features through serial heterogeneous convolution and wavelet convolution, after splicing, compression and attention screening, outputting through residual connection; multi-scale feature fusion; small target path feature enhancement fusion: extracting detail and context information through channel segmentation and double-branch parallel convolution; target classification and positioning. Through serial heterogeneous convolution and wavelet convolution and active fusion screening mechanism, the present embodiment significantly improves the diversity of basic features while reducing the amount of calculation; by increasing an independent pooling branch, global statistics and spatial guidance information are explicitly introduced, breaking through the limitation of traditional double-branch FPN / PAN or BiFPN which only optimizes the fusion weight between convolution features, achieving more sufficient and robust feature fusion with low computational cost; adopting a branch processing strategy, the quality of the input features of the detection head is strengthened, achieving a balance between model precision and speed.

The invention relates to a resource-pooling system and to a corresponding method for risk sharing of a variable number of risk exposure components. The risk exposure components are connected to the resource-pooling system by means of a plurality of payment receiving modules configured to receive and store payments from risk exposure components for the pooling of their risks. The total risk of the pooled risk exposure components comprises a first risk contribution associated to risk exposure in relation to loan losses, and a second risk contribution associated to risk exposure based on emergency expenses. The pooled risk is divided in a parameterizable risk part and a non-parameterizable risk part by means of an indexing module. In case of triggering a loss by means of a trigger module, the suffered loss is covered by releasing associated loans and emergency expenses of the risk exposure components.

The present disclosure relates to methods and apparatus for computing pooling operations on a GPU architecture. An example apparatus comprises: interface circuitry; machine readable instructions; and at least one processor circuitry programmed by the machine readable instructions to: allocate a computation task among one or more cores based on a batch dimension; perform a row pooling operation based on the batch dimension and a first pooling size to generate a first intermediate data output; perform a column pooling operation based on the batch dimension and a second pooling size to produce a second intermediate data output; and generate a final output based on the first intermediate data output and the second intermediate data output.

The invention provides a power equipment detection method and related equipment. The method comprises the following steps: processing acquired multi-modal data to obtain a compression tensor; performing power equipment fault prediction based on the compression tensor to obtain a fault level and a corresponding confidence coefficient; and in response to the condition that the confidence coefficient is smaller than a judgment threshold value, optimizing the detection parameters of the multi-modal data, obtaining the multi-modal data again based on the optimized detection parameters, and performing fault prediction on the power equipment again until the confidence coefficient is greater than or equal to the judgment threshold value. According to the embodiment of the invention, multi-modal data fusion and a dense connection network structure are combined with an optimization algorithm, key feature extraction is enhanced by utilizing a channel attention mechanism and mixed pooling, and detection parameters are dynamically adjusted by combining cuckoo, naked mole and grey wolf optimization algorithms, so that the detection robustness is improved, the consumption of computing resources is reduced, optimization convergence is accelerated, and the detection efficiency is improved. And a comprehensive and accurate quantitative basis is provided for power equipment fault diagnosis.

This invention relates to the fields of computer vision and machine learning, specifically to a lightweight classification and segmentation method and system for LiDAR point clouds based on GRSNet. The method includes: acquiring LiDAR point cloud data; generating a representative point set using a sampling method based on the golden ratio to reduce preprocessing complexity; inputting the point set into a GRSNet network for feature extraction, which consists of multiple processing stages. Each stage constructs a local region through a Sample_and_Group module, uses two SA_Ghost Block modules to learn shared weights and extract deep aggregation features, and performs feature aggregation through a Mixed Pooling module with randomly selected pooling methods; finally, classification or segmentation is completed based on the extracted features. This invention, through innovative sampling methods, a lightweight dual-path feature extraction module, and a randomized mixed pooling mechanism, significantly reduces the number of model parameters and computational complexity while maintaining high-precision classification and segmentation performance.

The invention discloses an intelligent process generation method and device based on knowledge graph embedding similarity matching, and belongs to the technical field of intelligent manufacturing process planning, and the method comprises the steps: constructing an assembly process knowledge graph; an assembly process sub-graph is extracted with the process as the center and serialized; the method comprises the following steps of: obtaining node-level embedding through a twin Transform encoder with shared parameters; obtaining graph-level embedding through batch embedding splicing and a pooling layer; performing bilinear interaction on an assembly subject, a relationship and an object by using an assembly element interaction neural tensor network, and extracting deep semantic features; outputting a similarity score of the two sub-graphs; retrieving the most similar process node based on the score; and executing shortest path search on the directed acyclic graph to complement missing nodes, and generating a target assembly process path. According to the method, rapid matching and automatic generation of different and semantically similar processes are realized, and the process reuse rate and the generation efficiency in a multi-variety, small-batch and rapid iterative manufacturing environment are remarkably improved.

The application provides a sequence prediction method and system of a pooling echo state network based on compressed sensing, belongs to the technical field of network information prediction, acquires sequence data required by a prediction task, constructs a pooling echo state network model based on compressed sensing, trains the network model, inputs the acquired sequence data into the trained pooling echo state network model, and obtains a prediction result; a pooling layer and a compressed sensing layer are added to a reservoir pool of the pooling echo state network model, the pooling layer is used for readjusting the weight of the state of the reservoir pool node, and the compressed sensing layer is used for sparse transformation and random subsampling of the node; based on the mechanism of compressed sensing and the pooling algorithm, the application provides a mechanism which can effectively reduce redundant nodes and improve the active performance of the nodes, effectively improves the model performance of the ESN, and makes the reservoir pool active while reducing the calculation amount, and improves the accuracy and operation efficiency of model calculation.

The invention provides a query execution time prediction method and system based on operator rate perception of a graph database execution plan, and relates to the technical field of database query optimization and intelligent data management. According to the method, firstly, an execution plan of target query is modeled into a directed acyclic graph, and characteristics such as operator operation types, line numbers and hit times are extracted; performing robust calibration and normalization through topological sequence aggregation update and combining with historical logs or small sample pilot run estimation operator rate factors, and then performing injection node update and graph-level pooling to obtain a plan representation vector; meanwhile, carrying out characterization coding on the query semantics to obtain a query representation vector; interactive fusion of plans and query representation is realized by utilizing an attention mechanism; and finally, inputting the fusion representation into the prediction network, and outputting the execution time in a set environment. According to the method, operator rate perception, semantic coding and attention interaction mechanisms are introduced, so that the accuracy and robustness of query execution time prediction are effectively improved.

An information transmission method and apparatus. The information transmission method comprises: a first device determines resource requirement information, the resource requirement information being used for indicating a requirement of the first device for a required BM resource, and the resource requirement information comprising one or more of the following: a real-time performance requirement, a pooling requirement, a sharing requirement, a security requirement, and a mobility requirement; the first device sends the resource requirement information to a second device; and the second device allocates to a BM resource instance a BM resource meeting the resource requirement information. Therefore, the BM resource allocated by the second device to the BM resource instance meets the requirement of the first device for the BM resource. The requirement indicated by the resource requirement information may be a requirement of a telecommunication cloud application for the BM resource. Therefore, the method facilitates the second device to deploy a BM resource instance meeting a requirement of the telecommunication cloud application for differentiation of BM resources.

The invention discloses an employment service management method and system based on an AI large model, and relates to the technical field of human resource management, and the method comprises the steps: integrating the semantic embedding of a semantic enhancement substring in a clustering result through weighted average pooling, generating a comprehensive semantic embedding vector of the semantic enhancement substring, and carrying out the semantic embedding of the semantic enhancement substring; calculating semantic similarity of the semantic embedding vector by using cosine similarity, and generating a recessive skill vector; calculating intensity similarity and semantic similarity on the basis of the normalized recessive skill vector, calculating comprehensive matching degree on the basis of the intensity similarity and the semantic similarity, converting the comprehensive matching degree into a cost value by using linear transformation, and solving and generating a pairing set by using a Hungary algorithm. According to the method, key semantic information in job seeker resumes and post description is enhanced by combining a semantic embedding technology and a substring clustering algorithm, and deep semantic understanding of a traditional text matching method is improved by combining semantic feature extraction with weighted average pooling.

The embodiment of the invention provides a vehicle-mounted intelligent driving platform 3D rendering method and system based on an unreal engine, electronic equipment and a storage medium, and the method comprises the steps: monitoring a vehicle someip message through a someip SDK, and analyzing ADAS core data (including a high-precision road network, a perception target and the like) through a protobuf protocol; converting the analysis data into a format capable of being recognized by an engine, and classifying the analysis data into corresponding data sets according to rendering scene logic after association caching; a standardized interface is called, rendering is executed by a lane rendering module, a perceptual object rendering module and the like, finally the rendering is packaged and published to an instrument terminal, a central control terminal and an HUD terminal, and multiple terminals are adapted only through configuration and UI layout. The problem of cross-terminal development fragmentation of a traditional HMI system is solved, one set of code resources is adaptive to multiple terminals, and cooperation and maintenance cost is reduced. By means of the real-time rendering capability of an unreal engine and a programmed grid body assembly, dynamic updating of road elements is achieved, and intelligent driving intuition is improved. An object pool technology is adopted to optimize memory management, and system stability is enhanced.