Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

98 results about "Linear network" patented technology

Multi-modal news recommendation method and system in combination with clock interests

The invention discloses a multi-modal news recommendation method and system in combination with clock interests, and relates to the technical field of data mining and recommendation methods.The method comprises the steps that a user behavior sequence, a candidate news set and a related timestamp set are extracted; obtaining the multi-modal coding representation of the historical interaction news through the time perception multi-modal feature coding; performing clock interest modeling based on multi-modal coding representation of historical interactive news, and fusing long and short-term interests through Gaussian weighted aggregation and long-term interest enhancement operation to obtain a user interest vector; and calculating the matching degree of the candidate news and the current user interest vector, applying time period sensitive suppression through time sequence gating, performing click probability prediction on the output of the time sequence gating by adopting a linear network, and performing a dynamic recommendation decision. According to the method, by means of the feature fusion technology of time context perception, the interest continuity of the hour granularity is captured, and the recommended content is more accurately matched with the requirements of the user in different time periods.
Owner:SOUTH CHINA AGRICULTURAL UNIVERSITY

Radiation source identification method based on attention mechanism and multi-domain feature fusion

The invention relates to a radiation source identification method based on attention mechanism and multi-domain feature fusion. The invention relates to a radiation source identification method based on attention mechanism and multi-domain feature fusion. The objective of the invention is to solve the problem that the existing method is poor in accuracy and reliability of specific radiation source identification. The method comprises the following steps: step 1, constructing a radiation source data set; step 2, constructing a radiation source identification network; the radiation source identification network comprises a variational mode decomposition (VMD) module, a discrete wavelet transform (DWT) module, a feature extractor, a multi-domain feature fusion module and a linear network; inputting the radiation source data constructed in the step 1 into a radiation source identification network, and outputting an identification result by the radiation source identification network; training the radiation source identification network to obtain a trained radiation source identification network; and step 3, inputting a radiation source to be detected into the trained radiation source identification network, and outputting an identification result by the trained radiation source identification network. The method is applied to the field of radiation source identification.
Owner:JILIN UNIVERSITY

Linear network coding for blockchains

ActiveUS12513012B1User identity/authority verificationAlgorithmAll-or-nothing transform
In a blockchain, transacting a record comprises performing linear network coding on the record or on a data file corresponding to the record, to produce a plurality of coded data parts; and storing at least one of the plurality of coded data parts on the blockchain or storing a proof of knowledge on the blockchain, the proof of knowledge derived from the coded data parts. Linear network coding coefficients might be derived from cryptographic hashes of the plurality of coded data parts. An all-or-nothing transform can use the cryptographic hashes to provide the proof of knowledge.
Owner:TYBALT LLC

Electromagnetic spectrum prediction method and system based on fractional Fourier transform

The invention discloses an electromagnetic spectrum prediction method and system based on fractional Fourier transform. The method comprises the following steps: constructing historical spectrum observation data in a target area; mapping the collected spectrum observation data to a corresponding fractional order Fourier domain by using adaptive fractional order Fourier transform, and adaptively adjusting fractional order parameters of a fractional order Fourier transform layer to output a fractional order Fourier domain containing predictable spectrum components; in the mapped fractional order Fourier domain, non-predictive noise in the converted data is filtered out; performing feature extraction and prediction on the frequency spectrum data of the fractional order Fourier domain after noise filtering by using a complex valued linear network to obtain a prediction result of the fractional order Fourier domain; and mapping the prediction result of the fractional Fourier domain back to the original time domain by adopting inverse fractional Fourier transform, and generating final frequency spectrum prediction output. According to the method, the accuracy of weak periodic spectrum prediction is effectively improved through fractional Fourier transform and a filtering strategy.
Owner:NANJING UNIV OF AERONAUTICS & ASTRONAUTICS

Accelerated simulation solving method and system for controllable self-recovery energy dissipation device

The invention provides an accelerated simulation solving method and system for a controllable self-recovery energy dissipation device, and the method comprises the steps: S1, enabling an inductance element between a nonlinear network and a linear network in a hybrid cascade DC power transmission system to be equivalent to a short transmission line, and building circuit models for the linear network and the nonlinear network respectively; s2, obtaining the value of the current controlled voltage source based on the node voltage of the previous simulation time step of the linear and nonlinear networks, the value of the controlled voltage source and the information interaction format; s3, node voltage solving is carried out based on the value of the controlled voltage source of the current simulation time step to obtain the node voltage of the current simulation time step, if the number of iterations is met, the operation is stopped, and otherwise, the step S2 is executed; the nonlinear network comprises a controllable self-recovery energy dissipation device; according to the invention, the energy dissipation device and the linear main circuit in the circuit system are decoupled in a sub-network manner, the calculation amount is reduced, and the iterative algorithm is used for solving, so that the high-efficiency and high-precision simulation of a nonlinear network and a linear network is ensured.
Owner:GLOBAL ENERGY INTERCONNECTION RES INST CO LTD

Chinese-blind conversion method based on consecutive character vector fusion Transform-GRU

The invention relates to a Chinese-blind conversion method of Transform-GRU based on consecutive character vector fusion, and belongs to the field of natural language processing and generation. According to the method, firstly, a word embedding technology for optimizing word segmentation continuous writing rules is realized through a multi-layer attention linear network, it is guaranteed that a model learns word vectors containing rich word segmentation rules, and compared with a general embedding technology, the method has higher pertinence for Chinese-blind translation; a Markov word segmentation model based on Gaussian normalization is designed, accurate word segmentation is achieved, and continuous word vectors are fused; according to the encoder design, high-level semantic information extraction is realized by an improved encoder formed by stacking a plurality of unidirectional extension layers and local attention layers, so that the problem of high complexity of a Transform network is improved, and targeted optimization is realized on Chinese and braille alphabets with consistent word orders; and the decoder designs a mask grouping self-attention mechanism and a GRU layer fused by bidirectional mutual attention gating, so that the capability of capturing rich semantics of coding information and the reasoning capability of the network can be improved.
Owner:BEIJING INST OF TECH +1

End-to-end throughput maximization method based on linear network

The invention provides an end-to-end throughput maximization method based on a linear network, and relates to the technical field of wireless sensor networks, and the method comprises the steps: firstly carrying out the linear deployment of sensor nodes in the network, and completing the intra-group and inter-group division of the sensor nodes; calculating the energy consumption of each node according to a transmission and receiving energy consumption model of the nodes to obtain the residual energy of the nodes; according to the residual energy of the nodes, the priorities of the nodes in the groups are calculated, the nodes in the groups distribute different backoff window waiting durations according to the priorities, and when the edge uplink nodes in the groups are ready for data transmission, the edge uplink nodes enter inter-group transmission competition; whether the data is within the data transmission interference range or not is judged through the inter-group interference model, after the backoff time is over, the source node carries out data transmission again, a retransmission mechanism is executed, and the reliability and the transmission success rate of the data are ensured.
Owner:SHENYANG UNIVERSITY OF TECHNOLOGY

Multi-modal self-adaptive preprocessing method for gas sensor based on Mamba state space model

The invention discloses a gas sensor multi-mode adaptive preprocessing method based on a Mama state space model, and relates to the field of gas detection, and the method comprises the following steps: synchronously collecting original signals, environmental parameters and historical time sequence data of a sensor; performing feature extraction on the three types of data to generate a sensor feature vector, an environment feature vector and a historical feature vector; the method comprises the following steps of: obtaining a final fusion feature through processing, mapping the final fusion feature to a 128-dimensional embedding space through a contrast learning encoder, inputting a three-layer residual error connection Mamba time sequence prediction network to output a zero offset, dynamically predicting process noise and observation noise through a neural network, executing adaptive Kalman filtering to output a smooth signal, and outputting a final fusion signal. Adopting a learnable piecewise linear network to carry out nonlinear correction on the filtering signal; the gas concentration value is calculated by integrating the correction signal and the zero offset, and finally a confidence score and a data quality mark are provided; according to the method, high-precision, real-time and robust pretreatment of various gas sensors in a complex environment is realized.
Owner:GUANGDONG COSCO SHIPPING HEAVY IND CO LTD

Cross-subject electroencephalogram emotion recognition method based on dynamic domain invariant representation decoupling and recombination

The invention discloses a cross-subject electroencephalogram emotion recognition method and system based on dynamic domain invariant representation decoupling and recombination, and belongs to the technical field of artificial intelligence. The invention provides a non-personalized decoupling and recombination framework for cross-subject electroencephalogram emotion recognition, and aims to separate emotion-related individual invariant features from cross-subject individual invariant features through complex electroencephalogram signal characterization obtained through dynamic decoupling, so that individual differences are eliminated while emotion classification performance is guaranteed. Specifically, the method comprises the following steps: carrying out original EEG data analysis and preprocessing by using MATLAB and Python MNE libraries; frequency spectrum and space features of EEG signals are extracted through a multi-channel frequency spectrum space self-attention mechanism module, and capture of emotional features is enhanced in combination with a self-attention mechanism and a cross-attention mechanism; a joint distribution alignment method based on a category prototype is adopted, and decoupled feature distribution is optimized, so that invariant features in subjects and invariant features among subjects have higher distinction degree in emotion classification; the optimized decoupling features are recombined through a linear network, the two features are coordinated to perform more sufficient emotion representation extraction, and emotion classification is performed through a multi-layer perceptron. According to the method, excellent cross-subject emotion recognition performance is obtained on multiple data sets, and the emotion recognition rate of the electroencephalogram signals in a cross-subject scene can be effectively improved.
Owner:NANJING UNIV OF AERONAUTICS & ASTRONAUTICS

Multi-modal dialogue emotion recognition method and device

The invention discloses a multi-modal dialogue emotion recognition method and device, and relates to the technical field of multi-modal emotion recognition. More accurate emotion prediction is realized through the steps of extracting multi-modal features, inferring an implicit relationship, constructing a mixed relationship graph, performing multi-modal semantic alignment and the like. The method comprises the steps of performing single feature vector extraction and fusion on an input multi-modal dialogue sequence needing emotion recognition to obtain an initial multi-modal feature vector, determining an implicit association relationship between sentences included in each modal according to a constructed similarity function, and performing emotion recognition on the multi-modal dialogue sequence according to the implicit association relationship. And mapping the representation subjected to weighted fusion with the visual and audio modal information to a word list through a linear network to obtain a predicted word generation probability of the target sentence, and determining a difference loss value between the predicted word probability and an actual word probability based on a cross entropy loss function to obtain a loss function.
Owner:NORTHWESTERN POLYTECHNICAL UNIV

Tranform-based near-field channel state information feedback method under super-large scale array

The invention discloses a near-field channel state information feedback method under a super-large scale array based on Transform, which comprises the following steps: in a downlink XL-MIMO system in a near-field area, a user side performs two-dimensional matrix processing on acquired CSI data and transmits the processed CSI data to a coding end; two-channel coding compression is carried out on the CSI matrix, a first channel uses an encoder to extract channel related characteristics, and a second channel enables an original CSI matrix to pass through a linear network so as to reserve original information; compressing the output of the two channels, and splicing to generate a low-dimensional code word to be fed back; the base station side decoding end splits the low-dimensional code word and decompresses the low-dimensional code word of the corresponding channel; decompressed dual-channel code words are respectively input to two ends of a decoder; and carrying out size change on the output of the decoder to obtain the estimated original CSI. According to the method, angle and distance information of a near-field channel is fully utilized, and efficient CSI compression and high-precision reconstruction are realized.
Owner:NANJING UNIV OF POSTS & TELECOMM

Ultra-short-term wind power prediction method and device based on improved Hilbert-Huang transform

The invention discloses an ultra-short-term wind power prediction method and device based on improved Hilbert-Huang transform, and relates to the field of wind power prediction.The method comprises the steps that collected wind power data and wind speed data are processed, and initial wind power data and initial wind speed data are obtained; adding the initial wind speed data into white noise, generating new wind speed data, and performing iterative empirical mode decomposition to obtain a plurality of mode components; sequentially subtracting the modal component from the initial wind speed data to obtain a residual component; converting the components through a Hilbert transform algorithm to obtain new sequence data; inputting the new sequence data into a linear network for processing to obtain corrected wind speed data; constructing an ultra-short-term wind power prediction model based on the sequential sequence lightweight adaptive network; and inputting the corrected wind speed data and the initial wind power data into a prediction model for prediction, and outputting ultra-short-term wind power prediction data. According to the invention, the predicted wind speed is closer to the real wind speed, and the power is predicted more accurately.
Owner:NORTHEAST DIANLI UNIVERSITY

Polynomial network coding

Vector network coding, such as linear network coding, is used to compute a plurality of coded data parts from original data, wherein each coded data part is computed from a cryptographic hash function of a previous coded data part. A recipient of the coded data parts can compute the cryptographic hash functions of the coded data to reproduce a system of linear equations, which can be solved to recover the original data. A cryptographic key that employs a polynomial over a finite field can have polynomial coefficients that comprise a function of vector network coding coefficients.
Owner:TYBALT LLC

Systems and methods for audio transport

According to disclosed embodiments, methods and systems of data transmission are provided. An aspect of the present disclosure is a method comprising receiving an audio stream, parsing the audio stream into packets, encoding each packet using Alphabet Linear Network Coding (ALNC), and transmitting the encoded packets.
Owner:DATAVAULT AI INC

Trim pass metadata prediction in video sequences using neural networks

Methods and systems for generating trim-pass metadata for high dynamic range (HDR) video are described. The trim-pass prediction pipeline includes a feature extraction network followed by a fully connected network which maps extracted features to trim-pass values. In a first architecture, the feature extraction network is based on four cascaded convolutional networks. In a second architecture, the feature extraction network is based on a modified MobileNetV3 neural network. In both architectures, the fully connected network is formed by a set of three linear networks, each set customized to best match its corresponding feature extraction network.
Owner:DOLBY LABORATORIES LICENSING CORP

Unmanned aerial vehicle image classification model generation method based on big and small model fusion

The invention discloses an unmanned aerial vehicle image classification model generation method based on big and small model fusion, and the method comprises the following steps: firstly, carrying out the preprocessing of data cleaning, positive and negative sampling and the like on the data collected by an unmanned aerial vehicle; designing a low-dimensional feature extraction module in the large model to extract low-dimensional Embedding, and ensuring the consistency of the dimension and the size of the Embedding; secondly, designing a high-dimensional feature extraction module to extract high-dimensional Embedding, and enabling the Embedding to have higher expression and discrimination capabilities through networks such as an attention mechanism and a linear network; the high-dimensional Embedding is input into a super network, and the super network uses the high-dimensional Embedding and the preprocessed data to generate an image classification small model for each unmanned aerial vehicle; and finally, through a picture classification result of the small model and the large model, analyzing the operation situation of the unmanned aerial vehicle and generating countermeasures. The method has the beneficial effects that the accuracy of unmanned aerial vehicle picture classification is improved, and professional countermeasures can be provided.
Owner:CHINA UNIV OF PETROLEUM (EAST CHINA)

Light quantum computing chip structure oriented to quantum neural network

The invention discloses a light quantum chip which is combined with high-dimensional coding, is provided with two data coding layers and two trainable entanglement layers, is compatible with a data recoding technology, and can be used for variable component subtasks such as a quantum neural network. The chip structure comprises a cascaded Mach-Zehnder interferometer binary tree array, d groups of photon pair sources, d groups of controlled unitary gate modules, a multi-quantum bit control Z gate, four d-dimensional adjustable linear networks and two state tomography linear networks. The cascaded MZI binary tree array and the controlled unitary group jointly form a first coding layer and a parameter-containing entanglement layer, and the adjustable optical linear network and the multi-quantum bit control Z gate jointly form a second coding layer and a parameter-containing entanglement layer. The two state chromatography linear networks can further improve the computing power of the chip in a mode of changing a measurement basis or executing quantum state chromatography. According to the method, the expression ability of the light quantum neural network is improved, and the development of a light quantum computing chip and the application of the light quantum computing chip in actual tasks are facilitated.
Owner:浙江大学宁波国际科创中心

Protocol for autoconfiguration of communication network

ActiveUS12701030B2Computer networkEngineering
A method for autoconfiguration of a plurality of nodes in a linear network allows extracting the address and position of each node. The method includes applying an identifier field for transmitting to the bus the bit sequence of the identifier of a chosen node. Then for at least for the first node to the last but one node, a field comprising a predetermined bit sequence is applied. The field comprises dominant bits, so a current is transmitted. Then, a further field is applied for transmitting any stored direction bit associated to that node and obtained in any previous iteration. The iteration continues by choosing a node different from a node chosen in any previous cycle, starting the communication, until all nodes are identified.
Owner:MELEXIS TECH NV

Time sequence prediction method based on frequency shift linear network

The invention discloses a time sequence prediction method based on a frequency shift linear network. The method comprises the following steps: S1, converting time sequence data from a time domain to a frequency domain; s2, performing segmentation processing on the frequency domain data in the S1 to obtain a plurality of continuous frequency band data; s3, performing frequency shift operation on the segmented frequency band data in the non-low-frequency region, and moving to a low-frequency region; s4, converting all frequency band data in the low-frequency region to a time domain; s5, respectively inputting the time domain data in the S4 into a plurality of trained linear networks to carry out time relation modeling; s6, converting a modeling result in S5 into a frequency domain; s7, restoring the frequency domain data in the step S6 to the original frequency band position; and S8, converting the frequency domain data restored in the S7 to a time domain to obtain a corresponding prediction result in the time domain. According to the method, the problem of low prediction accuracy caused by different learning capabilities for different frequency components in data in an existing time sequence prediction method based on deep learning is solved.
Owner:NO 15 INST OF CHINA ELECTRONICS TECH GRP

A Method for Calculating Nodal Electricity Price Considering Nonlinear Network Losses and Distributed Balancing Nodes

The present invention provides a method for calculating nodal electricity prices considering non-linear network losses and distributed balancing nodes, belonging to the field of electricity markets. The method of the present invention includes: establishing an improved DC optimal power flow model considering non-linear network losses; using a correction method for distributed balancing nodes of the load weight type to correct the nodal electricity price model; then, using a relaxation method based on the second-order cone to process the corrected nodal voltage model, solving the model and calculating the nodal electricity price to ensure the optimality of the electricity price result; finally, substituting the nodal electricity price and system parameters into the proposed sufficient conditions for exact relaxation to verify the rationality of the electricity price calculation result. The method of the present invention scientifically measures nodal electricity prices, adapts to the actual needs of rapid clearing and accurate solution in the spot market, helps to improve the enthusiasm of electricity users to participate in the spot market, helps to improve the resource allocation efficiency of the electricity spot market, and has practical value for the pricing mechanism of the electricity spot market in China.
Owner:XI AN JIAOTONG UNIV

Networked vehicle data driving switching gain state estimation method

The invention relates to a networked vehicle data-driven switching gain state estimation method. The method comprises the following steps: collecting system data of subsystems in each networked vehicle to construct a nonlinear network system as a state space model; collecting process data, and establishing a data-driven switching gain state observer; defining error data, and constructing an error system; setting stable operation conditions of the nonlinear network system according to the constant quantity and the gain matrix; and the nonlinear network system operates under a stable operation condition, so that the state estimation of the networked vehicle data-driven switching gain is realized. A large amount of system data of subsystems in each networked vehicle is collected to extract, insight, predict and emphasize the data volume and the data quality, so that the networked vehicle has the characteristics of self-adaption and flexibility; due to the fact that the data-driven switching gain state observer is established, real-time monitoring and switching gain control can be carried out on a network system, the efficiency and the quality of network data transmission are improved, and the running stability of the networked vehicle is guaranteed.
Owner:HAINAN UNIV

High-precision hybrid expert large model and fine tuning method thereof

The invention discloses a high-precision hybrid expert large model and a fine tuning method thereof. The hybrid expert large model is composed of a domain expert group and a shared expert group. A high-precision data transmission fine tuning module based on each linear network LN in the shared expert, wherein the fine tuning module comprises an extended linear network, a numerical value alignment network, a blocking controller and a precision converter; the extended linear network is of a replicated linear network (LN) structure and converts an input vector into a first data flow vector; the numerical value alignment network performs numerical value coding on the output data flow vector of the extended linear network to generate a second data flow vector; the blocking controller controls whether the extended linear network and the numerical alignment network participate in domain expert group reasoning or not according to binary data signals; the precision converter adjusts the precision of the data stream output by the numerical alignment network and the precision of the data of the shared expert linear network to be consistent; according to the invention, the flexibility and intelligent control capability of the high-precision hybrid expert large model can be enhanced.
Owner:TIANJIN UNIV

Selection of pivot positions for linear network codes

A method for encoding data includes: selecting a sequence of pivot candidate positions from a sequence of g encoded vectors to encode blocks of g data symbols in a round of encoding by the following steps: providing a set of g pivot candidate positions; selecting pivot candidate positions for the sequence from the set of pivot candidate positions; removing the selected pivot candidate positions from the set of pivot candidate positions; and repeating until the set of pivot candidate positions is empty and the sequence of selected pivot candidate positions in the round is non-linear. A set of encoded vectors is generated based on the sequence of selected pivot candidate positions, each encoded vector including zero-valued coefficients at positions within the encoded vector preceding the pivot candidate positions and non-zero-valued coefficients at least at the pivot candidate positions.
Owner:STANWULF CO

Tuning method for radio frequency matcher

PendingCN120433753AMultiple-port networksFrequency control by mechanical meansCurrent loadCapacitance
The invention discloses a tuning method for a radio frequency matcher, and relates to the technical field of radio frequency matchers. The method comprises the steps that an a coefficient value is written into a matcher program, and a matcher runs; the method comprises the following steps: reading a # imgabs0 # value, a # imgabs1 # value and a Y0 value in a linear network through a sensor; # imgabs2 #, # imgabs3 # and Y0 are substituted into a substitution formula to obtain an estimated value X0 of the current load impedance, and target capacitance # imgabs4 # tar and target capacitance # imgabs5 # tar are calculated according to the estimated value X0 in order to enable the impedance of the power supply end to reach a set value Ytar; the matcher controls a motor of the executing mechanism to enable the capacitors # imgabs6 # and # imgabs7 # to move to the calculated target positions # imgabs8 # tar and # imgabs9 # tar, the executing mechanism works according to a calibration curve between a pre-stored capacitor impedance value and a mechanical position, and the method can be suitable for any linear matching network such as an L shape, a T shape, a Pi shape and the like.
Owner:ANHUI XIRONG ZHAOBO TECH CO LTD

Pruning stage metadata prediction in video sequence using neural network

Methods and systems for generating pruning phase metadata for high dynamic range (HDR) video are described. The pruning stage prediction pipeline includes a feature extraction network followed by a fully connected network that maps the extracted features to pruning stage values. In the first architecture, the feature extraction network is based on four cascaded convolutional networks. In the second architecture, the feature extraction network is based on a modified MobileNetV3 neural network. In both architectures, the fully connected network is formed from a set of three linear networks, each set being customized to optimally match its corresponding feature extraction network.
Owner:DOLBY LABORATORIES LICENSING CORP

Model compression method based on low-bit floating-point number and quantization after model training

The invention discloses a model compression method based on a low-bit floating-point number and quantization after model training. The model compression is suitable for a convolutional network and a linear network. And a user is supported to compress a high-precision model into a model represented by a 4-bit or 8-bit low-bit floating-point number. According to the method, model compression is carried out through a bit width set by a user, in a model quantization module layer, a configuration search algorithm is used for finding an optimal low-bit floating-point number quantization configuration and supporting a mixed quantization scheme with the same bit width, and floating-point numbers with the same bit width have various combinations of exponent bits and mantissa bits; under the scene that data calibration exists between model quantization module layers, a joint calibration optimization process is used, output of an original model serves as a data calibration object of a current quantized model, output errors between the quantized model and the original model are reduced to the maximum extent, and therefore the precision of the compressed model is maintained. The model compression method for performing PTQ by using the low-bit floating-point number is provided, a new scheme is provided for PTQ, the method can be widely applied to quantization tasks after training of various model compression, and the precision after model compression is maintained.
Owner:BEIHANG UNIV

A Linear Network Coding Method Based on BLS

The present invention relates to a linear network coding method based on BLS. The source node divides the file to be signed into several file blocks and represents them as a prime domain #imgabs0# k dimensional vectors, which are then augmented to become a set of linearly independent vectors; a subspace generated based on this set of vectors is signed, and a label for the subspace and a signature for each basis vector are generated. The signatures are loaded into a data packet and sent to a downstream node; after receiving the data packet, the intermediate router or the destination node verifies the signature of the data packet from the same source node; the intermediate router linearly combines all verified data packets to form a new data packet and sends it to the downstream node; if the destination node receives a predetermined number of verified data packets, it can parse the original data sent by the source node. The present invention can not only resist pollution attacks in network coding and ensure data integrity and authenticity, but also reduce user computing complexity and ensure secure and efficient data transmission.
Owner:FUJIAN NORMAL UNIV

Compression index high-efficiency transmission verifiable encryption integration method for mass data

The application discloses a kind of mass data-oriented compressed index efficient transmission verifiable encryption integrated method, by segmented Elias-Fano compressed inverted list, RRR bitmap and rANS entropy coding are constructed into compressed index structure that can be directly queried, adaptive block is carried out with α-β link model through network interface, realizes flexible transmission using RaptorQ fountain code and random linear network coding, using encryption coprocessor is encrypted with AEAD-SIV mode and establishes Merkle / KZG verifiable structure, in combination with SE-Bloom supports search.The application can realize index, transmission, encryption end-to-end cooperation, reduce bit cost and tail delay, guarantee integrity, tamper resistance and privacy.
Owner:NAT COMPUTER NETWORK & INFORMATION SECURITY MANAGEMENT CENT GUANGXI BRANCH

A multi-modal news recommendation method and system combined with clock interest

This invention discloses a multimodal news recommendation method and system that incorporates clock-based interests, belonging to the field of data mining and recommendation methods. The method includes extracting user behavior sequences, candidate news sets, and related timestamp sets; obtaining multimodal encoded representations of historical interactive news through time-aware multimodal feature encoding; performing clock-based interest modeling based on these representations, and fusing long-term and short-term interests through Gaussian weighted aggregation and long-term interest enhancement operations to obtain user interest vectors; calculating the matching degree between candidate news and the current user interest vector, and applying time-sensitive suppression through time-series gating; using a linear network to predict click probabilities for dynamic recommendation decisions. This invention leverages time-context-aware feature fusion technology to capture hourly-level interest continuity, enabling more accurate matching of recommended content with user needs at different times.
Owner:SOUTH CHINA AGRICULTURAL UNIVERSITY

A distribution network voltage transformer state evaluation method and device

The present invention relates to the technical field of online monitoring of electric power metering, and specifically to a method and device for evaluating the state of a distribution network voltage transformer, the method comprising the following steps: segmenting endogenous data and then embedding variables, using a joint self-attention mechanism and fusing it with itself to input a linear network layer to obtain endogenous intermediate features, using a linear network layer to embed variables in exogenous data to obtain exogenous intermediate features respectively; applying a cross-attention mechanism to process the endogenous and exogenous intermediate features, and inputting the results into an MLP network to obtain a predicted value of the actual transformer ratio; performing multi-task learning on the predicted value of the actual transformer ratio to obtain the actual transformer ratio; calculating the current actual error of the transformer according to the actual transformer ratio combined with the rated voltage on the high-voltage side of the transformer to determine whether the current error state of the voltage transformer is abnormal. It can solve the problem that power-off monitoring is required in the prior art, which will lead to invalid maintenance or increase the power outage time.
Owner:WUHAN HANYANG POWER SUPPLY POWER ENG INSTALLATION TEAM