33results about How to "Reduce data size" patented technology

The invention discloses a power load prediction method based on frequency domain feature clustering and weighted fusion, which comprises the following steps: decomposing a power load signal by using Fourier transform, extracting frequency domain features of different frequency components, then clustering the frequency domain features, and independently establishing a personalized prediction model for each clustering category. During prediction, a plurality of models most similar to a target prediction sequence are selected, and prediction results of the models are fused in a weighted average mode. Compared with the existing power load prediction method, the frequency feature extraction method can better capture the potential important periodic features of the power load; user load data with similar frequency domain characteristics are clustered, so that different power utilization modes can be better identified; and through weighted fusion of prediction results of different models, the error of a single model is effectively reduced, and the resistance to noise is improved.

The invention discloses a shop commodity recommendation method and device, equipment and a medium, and the method comprises the steps: obtaining a search text inputted by a user in response to a search request of the user; extracting a keyword of the search text, and determining a corresponding category according to the search text; in the first stage, shop level recall operation is executed on the shop feature library, and part of shops with shop features matched with the keywords and the category to which the shops belong are recalled to form a shop recommendation list; and in the second stage, performing a commodity level recall operation on the commodity feature data of the commodities of the shops in the shop recommendation list in the commodity feature library, and recalling a part of commodities of which the commodity features are matched with the keywords and the categories to which the commodities belong, so as to form a commodity recommendation list. According to the method, the system pressure is effectively reduced through two stages of level recall operations.

The invention provides a call service processing method, which comprises the following steps: when a central user plane function network element center UPF receives a service message initiated by user equipment UE, identifying the service message meeting a preset condition to obtain a target service message; the center UPF reports the target service message to a session management network element SMF; and the SMF receives a target service message reported by the center UPF, and selects a local UPF to process the target service message. The invention further provides a call service processing system and device and a computer readable storage medium. According to the technical scheme of the invention, the load of an IP multimedia system IMS and a large network user plane function element UPF is reduced, and the new voice over air (VoNR +) call effect is improved.

The invention discloses a metal welding spot form detection method, and relates to the field of welding form identification, and the method comprises the following steps: constructing a sampling system for multi-angle progressive and multi-angle welding spot cloud collection of welding spots; firstly, a main point cloud is established at a main view angle, an algorithm is designed to judge the quality of the main point cloud, if data of the main point cloud is insufficient, auxiliary point clouds continue to be collected from other view angles, and then secondary judgment is carried out; designing a rapid 3D point cloud fusion scheme according to the set features of the to-be-detected sample; and finally, outputting the form description of the welding spot through machine learning. The invention further discloses a system for detecting and identifying the quality after welding spot welding, and the system is based on the method, the output welding spot form information is matched with the standard qualified welding spot, and the quality evaluation result of the welding spot is given in real time. The method is high in collection efficiency, can effectively overcome the environmental influence and the interference of sensor noise, improves the accuracy and integrity of the point cloud file, and improves the accuracy of the neural network recognition data.

The present invention discloses a method for extracting feature points based on time template matching, and belongs to the field of information security technology. The present invention includes performing block variance processing on the collected electromagnetic traces; performing digital-to-analog conversion processing and de-jitter processing on the electromagnetic traces after the block variance processing; performing Gaussian blur processing on the electromagnetic traces after the de-jitter processing, and performing digital-to-analog conversion processing and de-jitter processing again after the Gaussian blur is completed; intercepting sampling points from the electromagnetic traces to mark POIs; clustering POIs using the marking results, and performing extreme point processing; performing time template matching processing according to the POI time template and the number of categories. The feature point extraction method based on time template matching of the present invention can pre-process the electromagnetic traces collected during the operation of the dual-interface commercial smart chip card, and can significantly reduce the number of sampling points in the electromagnetic traces while retaining the POIs.

The invention discloses a method for generating an annual output time sequence of renewable energy, which belongs to the technical field of power system control, and comprises the following steps of: modeling a current residual component and a current trend component passing stability verification in a historical output time sequence by using an Irai random process; obtaining an initial simulation trend component and an initial simulation residual component; performing inverse difference processing on the periodic component and the target simulation trend component to obtain a target simulation trend component and a target simulation residual component, and finally reconstructing a target output time sequence by utilizing the target simulation trend component, the target simulation residual component and the periodic component. According to the scheme, mathematical modeling is carried out on the aperiodic component through the Irai random process, massive simulation curves can be generated, random features and time sequence correlation of the aperiodic component can be accurately reflected at the same time, key features of renewable energy under a long time scale can be accurately quantified, and the new energy consumption capability of a power grid can be accurately grasped according to the key features.

The invention relates to the technical field of dissolved oxygen prediction, and provides a dissolved oxygen prediction method based on data decomposition and related equipment, and the method comprises the steps: obtaining a plurality of water quality indexes of a target water body, and determining a plurality of target water quality indexes from all the water quality indexes; according to all the target water quality indexes, generating a dissolved oxygen data sequence of the target water body, and decomposing the dissolved oxygen data sequence to obtain a trend component and a seasonal component; predicting the trend component by using a trend prediction model to obtain a first dissolved oxygen predicted value, and predicting the seasonal component by using a seasonal prediction model to obtain a second dissolved oxygen predicted value; and fusing the first dissolved oxygen predicted value and the second dissolved oxygen predicted value to obtain a final dissolved oxygen predicted value of the target water body. According to the method, the dissolved oxygen prediction accuracy can be improved.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for identifying labels for a data set without disclosing the data set to any individual computing system. A method can include receiving, by a first computing system in a multi-party computation (MPC) system, a query including a first share and a second share of a given user profile. The second share is encrypted with a key that prevents the first computing system from accessing the second share. The second share is transmitted to a second computing system in the MPC system. The first computing system and the second computing system generate a machine learning model and identify respective first and second labels. The first computing system receives the second label from the second computing system as a response. The first computing system responds to the query with a response including the first label and the second label.

This invention provides a method and system for acoustic data compression and transmission for marine buoys, comprising the following steps: fusing floating-point data output from multiple channels into a single data stream; dividing all data in the single data stream into M frames of fixed length (N data points per frame), and performing data compression processing on each data point in each frame to compress the floating-point value of each data point into a three-digit integer code; encoding all data points in each frame to form a data packet, the single data stream forming M data packets, and transmitting the M data packets back to the ground station via satellite; this invention has the following advantages: by optimizing the data encoding and compression mechanism, this invention significantly reduces the data size while maintaining key acoustic characteristics, enabling the processed acoustic signal amplitude data to be reliably output under limited serial communication and low-bandwidth satellite channel conditions, thereby improving the availability and data transmission efficiency of the marine acoustic monitoring system.

The invention provides a network node processing method and device, electronic equipment and a computer program product, belongs to the field of network security, and aims to solve the problem that the existing method is difficult to quickly sense and respond to network security threats. The method comprises the following steps: preprocessing each node in a network, and determining candidate network nodes; identifying and evaluating network nodes which are not preprocessed in the network, and determining supplementary network nodes; taking the candidate network node and the supplementary network node as target network nodes; and obtaining target data of the target network node, analyzing the target data, and determining a network security risk according to an analysis result.

The present disclosure provides a data reconstruction method based on Radon domain sparse representation and related equipment, relating to the technical field of communication. The method comprises: obtaining low-resolution Radon coefficients; inputting the low-resolution Radon coefficients into a neural network of an encoding-decoding structure to obtain first high-resolution Radon coefficients; and increasing the sparsity of the first high-resolution Radon coefficients through an adaptive soft threshold function cascaded at the back end of the neural network to obtain second high-resolution Radon coefficients. The method can reduce the data size, operation time, and operation cost of time-domain inversion of reduced-time variable Radon transform, and improve the stability and resolution of reconstructed data.

The present invention discloses a method for three-dimensional point cloud target segmentation of a multi-level highway based on deep learning, and belongs to the technical field of three-dimensional point cloud target segmentation. The present invention first constructs a mapping relationship between preprocessed points and original point clouds through a point cloud data preprocessing operation based on a Map data structure and a K-D tree, and performs target segmentation processing on the preprocessed data based on the set multi-level highway point cloud target segmentation network, and then restores it to the original point cloud target segmentation result based on the mapping relationship. The data preprocessing operation of the present invention significantly reduces the amount of calculation. The target segmentation network of the present invention utilizes a feature aggregation module and an attention mechanism pooling to retain the geometric features of the point cloud data, and enhances the feature vector of the point, which is beneficial for the random sampling layer to extract global features. Establishing a residual network avoids gradient vanishing, and expands the receptive field of each point, which is beneficial for the random sampling layer to extract local geometric features of the point cloud, and greatly improves the accuracy of point cloud target segmentation.

The present invention discloses a high-throughput virtual screening method based on deep active learning, in which each ASCII character of a SMILES string of a compound molecule is converted into a corresponding number as its starting encoding vector, and then a deep learning model is used to automatically learn the implicit features of the compound in the SMILES string. In combination with active learning technology, the amount of data required for annotation for model training is greatly reduced. Only 0.1% to 0.5% of the samples in the entire compound library need to be annotated, which can greatly reduce the time and training cost of annotating data sets and model training. In addition, the present invention adopts a deep learning model as a virtual screening engine, and the virtual screening accuracy is high.

The invention discloses a large language model quantification method based on wavelet transform and BiLLM. The method comprises the following steps: inputting LLM; for each network weight matrix of the LLM, applying Haar wavelet transform to decompose the network weight matrix to obtain a wavelet coefficient matrix; partitioning a low-frequency coefficient part in the wavelet coefficient matrix, selecting an optimal threshold value, setting the weight smaller than the optimal threshold value to be zero, quantizing the part not smaller than the optimal threshold value, and compressing the low-frequency coefficient part of the wavelet coefficient matrix; compressing a high-frequency coefficient part of the wavelet coefficient matrix by applying a BiLLM binary residual approximation method; after each network weight matrix of the LLM is processed, quantization of the LLM is completed; wherein the quantization threshold value of the wavelet low-frequency coefficient adopts linear quantization in a floating-point number index range to replace linear quantization on a real number axis. According to the method, the traditional wavelet transform data compression strategy and the LLM-oriented BiLLM binary residual approximation algorithm are combined, and the LLM compression storage is realized on the basis of keeping the LLM correctness.

The invention discloses a remote upgrade data transmission method, device and equipment, and relates to the technical field of remote upgrade.The method comprises the steps that according to a remote upgrade request of to-be-upgraded equipment, an incremental upgrade package of the to-be-upgraded equipment is obtained; determining a compression algorithm set corresponding to the incremental upgrade package according to the data type of the data in the incremental upgrade package; the compression algorithm set comprises a complex compression algorithm corresponding to each data type; compressing the incremental upgrade package by using a complex compression algorithm in a compression algorithm set to obtain an upgrade compressed package; wirelessly transmitting the upgrading compressed package to the to-be-upgraded equipment so as to enable the to-be-upgraded equipment to complete upgrading by using the upgrading compressed package; according to the method, the complex compression algorithms corresponding to the data types are adaptively selected and combined for use according to the data types of the data in the incremental upgrade package, so that the compression effect of the incremental upgrade package is ensured, the data size of the wirelessly transmitted upgrade package is reduced, and the remote upgrade time is shortened.

The present invention proposes a network data transmission method and device, which relates to the technical field of data transmission. The method comprises: sending the data to be transmitted to a proxy cache server to generate a cache copy. Obtaining and placing the cache copy into a cache unit according to the data sensitivity level of the data to be transmitted. Obtaining the association relationship between the data to be transmitted corresponding to all cache copies in the cache unit, and determining the network construction method according to the association relationship. Constructing a data network according to the network construction method. According to a preset method, aggregated data is obtained, and according to the PEGASIS protocol, the aggregated data is made into a data packet and transmitted to a receiver. By reducing transmission energy consumption and improving data transmission efficiency and accuracy through data aggregation, the purpose of increasing the data transmission speed while ensuring accurate data transmission is achieved. The receiver arranges all data packets according to the metadata of each data packet and the data network to obtain an arrangement result.

The invention discloses an image projection method and system for interconnection between a mobile phone and a vehicle-mounted infotainment system, and the method enables the content of data transmission to be smaller and the transmission rate to be faster under the condition that the data size of unnecessary transition images is reduced through compressing the definition and quality of the transition images between a first target image and a second target image. The delay between the mobile phone and the vehicle-mounted terminal is reduced, so that the interconnection synchronism of the mobile phone and the vehicle-mounted terminal is improved, and the operation of remote screen projection is smoother; and an adaptive compression algorithm is utilized, upper and lower threshold values of a network data transmission speed are set, and an optimal compression mode is automatically selected, so that the data transmission of handcart interconnection has an intelligent adjustment function, and better image quality and use experience are kept in the optimal compression mode under the condition of keeping smooth screen projection of interconnection between a vehicle machine and a mobile phone.

The invention discloses a data networking terminal unbalanced data intersection method and a related device, and the method comprises the steps: screening the data of a first data party, and obtaining a sample data set; sending the sample data set to a second data party, and receiving hit information and a second data set returned by the second data party; sending hit information to the first data party, and receiving a first data set returned by the first data party; and carrying out intersection on the first data set and the second data set. The data of the first data party and the data of the second data party are simplified before intersection, the first data set and the second data set are obtained, furthermore, intersection is conducted on the first data set and the second data set, the scale of data needing to be processed in the intersection process is greatly reduced, and the efficiency of unbalanced data intersection is improved.

The application provides an image processing method, device and equipment and a computer readable storage medium. The method comprises: obtaining an input image to be processed and natural language reference text; mapping the input image to a visual token set and mapping the natural language reference text to a text token set; filtering the visual token set based on each text token in the text token set to obtain a candidate visual token subset corresponding to each text token; performing semantic analysis on the natural language reference text to obtain a logical constraint condition; performing alignment processing on the text token in the text token set and the visual token in the candidate visual token subset based on the logical constraint condition to obtain a mapping relationship; and determining target position information of the natural language reference text in the input image based on the mapping relationship. Through the application, the redundant calculation overhead of cross-modal interaction can be reduced, fine-grained matching can be guided by the logical constraint, and the accuracy of semantic alignment between the text and the image can be improved.

The present application provides a link tracking method, device, equipment and computer-readable storage medium, which are used to effectively reduce the data scale of link tracking processing and reduce data processing costs. It is particularly suitable for debugging applications, so that staff can directly locate and optimize bugs based on tracking data, significantly improving inspection efficiency. A link tracking method provided by the present application includes: a server receives an interception request initiated by a UE, wherein the interception request is used to request the server to intercept a target user request initiated by the UE, and the target user request is a user request that the UE confirms as an abnormal request; the server intercepts the target user request and performs link tracking processing on the target user request, wherein the link tracking processing is used to track the initiation trajectory of the target user request; the server stores the tracking data obtained by the link tracking processing.

The invention relates to the technical field of carbon emission monitoring, and discloses a carbon emission monitoring and early warning analysis system based on edge calculation. The system comprises a dynamic topology mapping module which is used for establishing dynamic topology mapping of an emission source in a monitoring area and recording a state transition path; the self-adaptive probe deployment module is used for deploying a probe associated with the monitoring dimension and the path according to the mapping at the edge node; the early warning identification module is used for comparing the time slice sequence with a path identification early warning point; the evolution trajectory generation module is used for generating a parameter evolution trajectory; the stage analysis and traceability module is used for distinguishing a steady state stage from a transient stage and executing inertia and sudden change analysis; and the autonomous optimization module is used for constructing a double-layer parameter correction model and driving the monitoring process to be optimized.

The present disclosure describes systems and techniques for controlling access to user information using transient user identifiers. In one aspect, a method includes determining, for a given domain, a user's engagement with content provided by the given domain for display by an application at a client device of the user. Based on the user's engagement, a link between a user identifier of a user extending the application for the given domain is determined. In response to determining a link between the user identifier of the user extending the application for the given domain, one or more future domain-specific transient user identifiers for the user and the given domain are obtained. An attestation record including a current domain-specific transient user identifier and one or more future domain-specific user identifiers is generated, and the attestation record is sent to the given domain.

The present invention discloses a run-length encoding accelerator and method for a sparse CNN neural network model. The accelerator includes: a top-level controller, which is used to identify the input and output data types and perform group processing, and transmit weight data and excitation data to a run-length encoding module; a run-length encoding module, which is used to compress the result data output by the calculation module, and transmit the compressed and encoded calculation results to an external memory; a run-length decoding module, which is used to decompress the data read from the external memory and transmit the data to a data gating; the data gating is used to identify the zero value of the input excitation data and the weight data, and skip the multiplication and addition operation for the zero value; and a calculation module, which is used to perform multiplication and addition operations on the weight data and excitation data transmitted by the data gating and give the calculation results. The present invention reduces the storage space and power consumption required for the CNN model based on data compression and data gating, and has the advantages of high energy efficiency and no need to access a large amount of memory.

The present disclosure provides a user information security risk assessment method and device, electronic equipment and medium, relates to the technical field of data processing, and particularly relates to the technical field of natural language processing, large language model and information security. The implementation scheme is: obtaining first page data of a first page of a target application program, wherein the first page is used to display a privacy policy describing a user information list of the target application program, and the first page includes at least one first link; processing the first page data using a large model, wherein the large model is used to: screening at least one second link from the at least one first link, wherein each second link in the at least one second link indicates a first sub-page used to display other content in addition to the privacy policy; and extracting an evaluation data list from the first page data excluding the at least one second link; and evaluating the degree of information security risk when a user uses the target application program according to the evaluation data list.

The present application relates to a method, device, computer equipment and storage medium for location data partitioning. The method comprises: obtaining location data to be partitioned; performing clustering processing on the location data, and obtaining candidate point data according to the clustering processing result; performing business partitioning processing on the candidate point data according to preset business constraints, and obtaining business units corresponding to the location data. The method can improve the partitioning processing effect of location data.

The application discloses a single station terminal fingerprint positioning method based on a square weighted K nearest neighbor algorithm and belongs to the technical field of fingerprint positioning. The method comprises the following steps: based on a channel frequency response (CFR) matrix, a position fingerprint matrix (FRCPM) with higher positioning precision is extracted; a fingerprint compression is performed on the fingerprint matrix (FRCPM) to obtain a compressed fingerprint (FCPV) with smaller data scale and being more beneficial to improving positioning speed; and a square weighted K nearest neighbor matching method with higher positioning precision is obtained by improving the weighted K nearest neighbor method.

The invention provides a contribution degree-based bad root cause analysis method and system, and relates to the technical field of bad root cause analysis, and the method comprises the steps: carrying out the defect rate calculation based on the preprocessed original processing data, so as to obtain the product defect rate; performing production path identification and production path combination based on the preprocessed original processing data to obtain product resume data; carrying out data splicing on the product reject ratio and the product resume data so as to construct the reject ratio under each production path combination; contribution rate calculation is carried out based on the reject ratio under each production path combination so as to obtain the contribution rate of each production path combination in the reject ratio; and sorting and screening the contribution rate of each production path combination in the reject ratio to obtain a suspicious bad root cause. By calculating the contribution rate, the influence degree of different production path combinations on the product badness in the whole machining process flow can be quantitatively evaluated, so that the potential process defect root cause can be identified more accurately.

The present invention discloses a rolling runoff forecasting method and system based on data drive, which divides historical runoff data and historical rainfall data according to sliding window size L and sliding window step size Step, constructs a sample database; obtains current runoff sequence R and current rainfall sequence F; uses current rainfall sequence F to perform similarity calculation in the sample database, obtains K samples with the highest similarity; and calculates runoff forecast value based on current runoff sequence R, current rainfall sequence F, and K samples with the highest similarity using a runoff forecasting model. The present invention can significantly reduce the amount of data involved in runoff forecasting, especially for the flood season and rainy season, improves the speed and accuracy of forecasting, and is of great significance for flood prevention and disaster reduction work.

Provided are a moving image reception device, a moving image transmission device, a moving image transmission / reception system, a control method, and a program capable of generating image data having an appropriate data size in consideration of a delay. An operation data transmission unit (44) transmits operation data corresponding to an input operation of a user to a cloud server (10). A VSP reception unit (46) receives a packet which is transmitted from the cloud server (10) in response to a start of generation of the frame image based on the operation data and which is associated with this operation data. A terminal-side traffic control unit (54) controls a data size of the image data to be subsequently transmitted by the cloud server (10), on the basis of a packet reception time being a time from a timing of the transmission of the operation data to a timing of the reception of the packet associated with this operation data.

The application discloses a three-dimensional point cloud semantic segmentation method, device, equipment, medium and product, and relates to the technical field of computer vision and three-dimensional point cloud processing. The method comprises the following steps: performing sparse voxelization processing on three-dimensional point cloud data to establish a point-to-voxel mapping relationship; performing geometric feature extraction on each non-empty voxel; adopting a clustering algorithm to aggregate voxel features, and determining a k-nearest neighbor graph based on the center coordinates of super points; performing parallel coding processing of feature extraction by adopting a double-flow coding strategy to obtain voxel flow local features and super point flow global features; projecting the super point flow global features back to the voxel dimension, and splicing and fusing the voxel flow local features; performing enhancement processing on the fused features based on a self-attention mechanism, splicing the attention features obtained after the enhancement processing with the fused features, and then performing semantic classification; and determining a prediction result of semantic segmentation according to the classification category. The application can improve segmentation accuracy while ensuring computing efficiency.