59 results about "Linear network" patented technology

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.

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.

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.

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.

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.

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.

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.

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.

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.

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.

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.

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.

This invention relates to a data-driven switching gain state estimation method for connected vehicles. The method includes: collecting system data from various subsystems within the connected 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 operating conditions for the nonlinear network system based on constant quantities and the gain matrix; and operating the nonlinear network system under stable conditions to achieve data-driven switching gain state estimation for the connected vehicle. By collecting a large amount of system data from various subsystems within the connected vehicle for extraction, insight, and prediction, the method emphasizes both data quantity and quality, enabling the connected vehicle to exhibit adaptive and flexible characteristics. Because a data-driven switching gain state observer is established, real-time monitoring and switching gain control of the network system can be performed to improve the efficiency and quality of network data transmission and ensure the stability of the connected vehicle's operation.

This invention discloses a method and system for optimizing data interaction between intelligent terminals based on link state awareness, relating to the field of wireless communication data transmission technology. The method includes: constructing a directed acyclic graph topology; performing cooperative topology state discovery to generate a real-time link state communication topology; performing hierarchical packetization for transmission quality constraints; performing random linear network coding to obtain multiple linearly independent coded packets; performing distributed congestion-aware scheduling decisions to obtain multiple primary transmission links; performing differential coding and obfuscation transformation to obtain multiple obfuscated coded packets; and concurrently transmitting multiple obfuscated coded packets to a second terminal for Gaussian elimination decoding and reception. This invention solves the technical problems of weak topology adaptability, unreasonable congestion scheduling, and insufficient security protection in existing technologies, achieving efficient and secure data interaction between terminals and improving the real-time performance and transmission security of local area network data transmission.

This application discloses a method, device, terminal, and medium for assessing the synchronous stability of grid-connected wind power. The solution obtains the topology information of the grid-connected wind power combined transmission system and constructs an equivalent circuit. Then, combining the preset grid-connected wind power control logic and the principle of linear network superposition, it constructs an output active power relationship under different control conditions, including the coupling relationship between the output active power of the grid-connected wind power and the terminal voltage and output phase angle of the grid-connected wind power. Based on this output active power relationship, it generates the output phase angle characteristic curve of the grid-connected wind power and calculates the output phase angle acceleration area and the maximum deceleration area. Finally, based on the area comparison results, it determines the synchronous stability assessment result of the grid-connected wind power, thereby accurately assessing the synchronous stability of the grid-connected wind power in the grid-connected wind power combined transmission system, improving the reliability of the assessment results, and ensuring the safe and stable operation of the new energy transmission system.

The invention provides a linear network and a relay dynamic adjustment method thereof, the method is applied to a communication front-end processor of the linear network, the linear network further comprises a plurality of acquisition stations connected in sequence, and at least one relay acquisition station is arranged in the plurality of acquisition stations; the communication front-end processor obtains signal quality margins of the relay acquisition stations, and the signal quality margins are matched with peak-to-peak value amplitude margins and signal-to-noise ratio margins of the relay acquisition stations; and performing relay dynamic adjustment according to the signal quality margin of the relay acquisition station and a preset safety threshold. Through obtaining the signal quality margin of the relay acquisition station, the change of the industrial environment is quantified, and then the relay dynamic adjustment is adaptively performed, so that the relay layout is optimized, the communication delay and energy consumption are reduced as much as possible, and the data transmission efficiency is ensured under the condition that the communication signal quality of the acquisition station in the linear network is ensured.

The invention discloses a wireless communication method and a wireless communication system. In the wireless communication method, each of N slave unmanned aerial vehicles passing an authentication program confirms that the slave unmanned aerial vehicles have the same time scale area as ground processing equipment according to a synchronization timestamp received from the ground processing equipment, selecting a corresponding key in a random key group permutation sequence as a current key according to a time point of receiving the random key group permutation sequence from the ground processing equipment, and encrypting k public advancing tracks received from the ground processing equipment by using the current key based on a random perturbation algorithm to obtain perturbation encryption data, transmitting the disturbance encrypted data to the main unmanned aerial vehicle according to the received serial number and coordinate of the main unmanned aerial vehicle from the ground processing equipment; and the master unmanned aerial vehicle executes random linear network coding on the received N disturbance encrypted data transmitted by the N slave unmanned aerial vehicles to generate a coding code block, and transmits the coding code block to ground processing equipment.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for selecting an action in response to each context in a sequence of contexts. One of these methods includes maintaining data specifying a respective gated linear network corresponding to each of a plurality of actions; for each context in a sequence of contexts: for each action, processing the context using the gated linear network corresponding to the action to generate a predicted probability; for each action, generating an action score for the action based on at least the predicted probability; and selecting an action to perform in response to the context based on the action scores.

The invention provides a procambarus clarkii individual identification method and system based on biological characteristics, and relates to the technical field of deep learning. Extracting local digital biological characteristics of each slice image and the target image and global digital biological characteristics of the target image, and fusing the local digital biological characteristics and the global digital biological characteristics to obtain a digital biological characteristic matrix; inputting the digital biological characteristic matrix into an improved graph neural network model, respectively generating a query and key through an independent multi-layer perceptron, calculating an edge weight through a zoom dot product of the query and key, and carrying out multi-head attention splicing to obtain a plurality of node characteristics; splicing different node features, generating an attention score for each node based on an aggregation module of a gating mechanism, and outputting a global representation vector of each group of point sets after global pooling operation; and mapping the global representation vector into a fixed dimension through a linear network, embedding the fixed dimension as a final point set, and finally outputting to obtain a procambarus clarkii individual identification result.

The present disclosure provides a method and system for identifying individual procambarus clarkii of biological characteristics, relating to the technical field of deep learning, comprising: extracting local and global digital biological characteristics of each slice image and a target image, and fusing the local and global digital biological characteristics to obtain a digital biological characteristic matrix; inputting the digital biological characteristic matrix into an improved graph neural network model, generating a query and a key through an independent multi-layer perceptron, calculating edge weights through a scaling dot product of the query-key, and obtaining multiple node features after multi-head attention splicing; splicing different node features, generating an attention score for each node based on a gating mechanism aggregation module, and outputting a global representation vector of each group of point sets after global pooling operation; mapping the global representation vector to a fixed dimension through a linear network, and taking it as a final point set embedding, and finally outputting a procambarus clarkii individual identification result.

The present application provides a kind of net zero energy consumption building radiant heating coil spacing prediction method, it is related to coil spacing prediction field.Process includes building net zero energy consumption building radiant heating coil spacing data set, each net zero energy consumption building data is built as multidimensional input sequence, high-dimensional feature sequence is generated using double-layer nonlinear transformation network, building thermal disturbance feature sequence is generated in combination with the mean and standard deviation of high-dimensional feature sequence, introduce thermal response significance weight value and carry out weighted fusion, obtain fusion characteristics after layer normalization, extract thermal potential value and thermal slope, and input nonlinear gate structure with fusion characteristics, to obtain main heat transfer characteristics and residual heat characteristics, generate fusion response weight using element-by-element multiplication, weighted sum of main heat transfer characteristics and residual heat characteristics to obtain weighted thermal response characteristics, finally, weighted thermal response characteristics are input into linear network, and output radiant heating coil spacing prediction value.This method can realize the intelligent prediction of coil spacing control parameters.

The application discloses a train network architecture based on Ethernet deep fusion, which comprises a red network switch topology, a blue network switch topology and train subsystems connected to the blue network switch topology and the red network switch topology; the train subsystems are redundantly designed and arranged on a head car, a middle car or a tail car of the train; the backbone network topologies of the red network and the blue network are not limited and are suitable for ring network topology, linear network topology or link aggregation network topology; the application saves switch hardware devices and communication cable materials; the whole train network data ports are centrally managed, so that traffic monitoring and fault positioning are facilitated; the application builds an intensive integrated network architecture and fully utilizes the data bearing capacity of a gigabit backbone network; and the application uniformly plans a train-ground safety transmission boundary, which is beneficial to the deployment and drilling of diversified safety strategies.

The application belongs to the technical field of power grid power supply, and particularly relates to a measurement method and system for tower grounding resistance, which comprises the following steps: constructing a chain linear network model, setting an engineering typical value of the grounding impedance of each tower, constructing a prior admittance matrix, inverting the prior admittance matrix to obtain a prior impedance matrix, injecting a mixed excitation current signal with multiple different frequencies into the grounding lead of the measured tower, synchronously collecting the measured voltage of all nodes, separating the measured voltage under each frequency, calculating an admittance deviation matrix and an impedance deviation matrix, establishing an approximate linear relationship between the impedance deviation matrix and the admittance deviation matrix and a voltage difference vector, solving the approximate linear relationship, obtaining the impedance deviation of the grounding impedance under each frequency relative to the engineering typical value of the grounding impedance, calculating the actual grounding impedance, and separating the actual grounding impedance to obtain the grounding resistance. The application realizes the inversion of the grounding resistance from the external port measurement value without disassembling any grounding lead.

The application discloses a preset performance synchronization control method of a nonlinear networked time-delay system based on state translation transformation, which transforms communication time delay of the system by using state translation transformation technology, designs a preset time synchronization control scheme through the system after translation transformation, analyzes the synchronization of the transformed system, indirectly obtains effectiveness of the original system under the preset time controller, and finally realizes preset time synchronization control of the nonlinear networked time-delay system. Based on the constructed Lyapunov function, sufficient conditions of preset time synchronization are obtained, under the conditions, even if the system is affected by communication time delay or preset performance index changes, the synchronization control of the nonlinear networked time-delay system can still be realized under the preset time requirement. Finally, the effectiveness of the synchronization control scheme is verified through numerical simulation.

This invention discloses an intelligent detection method for disinformation on social networks based on deep learning. The method includes: collecting text information of all messages on a social network over a given time period, processing reply text information and topological information, and performing text preprocessing; inputting the message text information into a local information encoder to output message text feature vectors; inputting the topological information of the social network topology graph formed by the messages into a global information encoder to output message topological information feature vectors; concatenating the message text feature vectors and topological information feature vectors to obtain a fused feature vector; and inputting the fused feature vector into a linear network to output the final prediction vector. This invention, using fixed-length samples while ensuring training efficiency, significantly enhances the model's feature extraction capability by fusing network message content, reply message content, and message topological information in the propagation graph, overcoming the problem of insufficient model feature extraction ability.

The application provides a linear network and a relay dynamic adjustment method thereof. The method is applied to a communication front end machine of the linear network. The linear network further comprises a plurality of acquisition stations connected in sequence, and at least one relay acquisition station is arranged in the plurality of acquisition stations. The communication front end machine acquires signal quality margins of each relay acquisition station. The signal quality margin is matched with a peak-to-peak amplitude margin and a signal-to-noise ratio margin of the relay acquisition station. According to the signal quality margin of the relay acquisition station and a preset safety threshold, relay dynamic adjustment is performed. The signal quality margin of the relay acquisition station is acquired to quantify the change of an industrial environment, and then the relay dynamic adjustment is adaptively performed. In the case of guaranteeing the communication signal quality of the acquisition stations in the linear network, the relay layout is optimized, the communication delay and the energy consumption are reduced as much as possible, and the data transmission efficiency is guaranteed.

The application discloses a multi-label text classification method and device, the method comprises the following steps: obtaining a plurality of texts of known label categories, constructing a training set and a test set; inputting each text in the training set and a plurality of label categories into a BERT model to output a first embedded sequence; converting the established relationship matrix into a second embedded sequence; inputting the first embedded sequence and the second embedded sequence into a relative attention network (RAT) to output semantic correlation information and a plurality of internal correlation information, and converting the semantic correlation information and the plurality of internal correlation information into a corresponding one-dimensional vector through a bidirectional LSTM network; mapping the one-dimensional vector to the plurality of label categories in the training set through a linear network to obtain a label category prediction result of each text; calculating a loss value of the network model, updating network model parameters according to the loss value to obtain a trained network model; and classifying to-be-classified texts by using the tested network model, so that the accuracy of multi-label text classification can be improved under low resource conditions.

The utility model discloses a novel multi-aperture linear network switch, and relates to the technical field of network switches, the periphery of the bottom of a network switch main body is provided with support rods, the bottom of each support rod is provided with a cushion block, and the rear end face of the network switch main body is provided with a wiring port. Heat dissipation grooves are formed in the left side and the right side of the top of the network switch main body; clamping grooves matched with the cushion blocks are formed in the periphery of the top of the network switch main body; when the indoor temperature is too high in summer, the cover plate can be opened through the hinge, the contact area between the inner cavity of the network switch main body and the outside is increased, and the working heat dissipation fan can increase the air flowing speed at the top of the network switch main body, so that the heat dissipation processing effect of the heat dissipation fan can be improved, and the heat dissipation effect is improved; when a plurality of groups of network switch main bodies are stacked together, gaps can be formed between the adjacent network switch main bodies through the arranged supporting rods, so that heat accumulation can be avoided, and heat dissipation is facilitated.

The invention discloses a data synchronization method and system based on medical consumables, and relates to the technical field of data synchronization, and the method comprises the steps: collecting monitoring target list data of medical equipment, carrying out the arrangement and structuring of the sampled data, forming an independent data group, extracting the collected data of different sensor data items, and forming an organization matrix; and carrying out linear coding on the organization matrix by using a random linear network coding (RLNC) method, generating a redundant coding packet data matrix by using matrix multiplication, and carrying out homomorphic encryption on coding packet data. According to the method, data protection and calculation operation are seamlessly combined through homomorphic encryption, encryption monitoring data can be directly processed, and encryption potential features are generated through combination of time random features and mapping hidden features, so that the hidden space features can express time dependence features and global features of random coding at the same time; and the multi-dimensional characteristic fusion capability in a medical consumable monitoring scene is enhanced.

The invention discloses a visual generative model training method and device based on a linear network layer, equipment, a medium and a product. Comprising the following steps: inputting a first training image into a subject hidden space for feature coding to obtain a training image hidden vector of a first spatial dimension, converting the training image hidden vector into a training image hidden vector of a second spatial dimension through a linear network layer, and inputting the training image hidden vector into a visual generation sub-model to obtain a generated image hidden vector of the second spatial dimension; converting the generated image implicit vector of the second spatial dimension into a generated image implicit vector of the first spatial dimension through a linear network layer, and then performing feature decoding through a subject implicit space to obtain a generated image; updating model parameters of the visual generation sub-model based on the loss value to obtain a target visual generation sub-model; and obtaining a vision generation model based on the target vision auto-encoder and the target vision generation sub-model. According to the invention, the quality of the generated image can be improved, the model training and image generation efficiency can be improved, and the computing power cost is reduced.