51 results about "Computation sharing" patented technology

Techniques for automated tagging of visual content are described. A pairwise model is used to encode images and videos with a vision encoder, and encode keywords and phrases from an interest graph with a text encoder. Similarity layers compare these cross-modality embeddings by calculating distance in a shared embedding space. Scores indicate the association between visual features and text. A pairwise loss function brings together matched pairs while separating non-matches during training. Scores exceeding a threshold tag content with relevant keywords and phrases. The pairwise architecture relates images and text despite limited associated text. It leverages vision-to-text understanding for accurate tagging without per-class labels. Contrastive similarity techniques associate visual patterns with textual concepts. Automated tagging organizes user-generated content by topics using this scalable cross-modality approach.

The invention discloses an intelligent network data security protection method, system and device and a medium thereof, and the method comprises the following steps: obtaining network flow data, monitoring the network flow of the network flow data through a flow collection device and a network monitoring tool, extracting the feature information of a data packet, carrying out the preprocessing based on the feature information of the data packet, and obtaining the network flow data; according to the convolutional neural network and the LSTM, time sequence features are captured, modeling is carried out according to the convolutional neural network, and a relation between devices in the network and a topological structure of a data stream are determined; on the basis of privacy protection and calculation sharing of homomorphic encryption, encryption calculation is carried out through a Paillier encryption algorithm, summation and multiplication operations are carried out on encrypted data, the encrypted data are distributed on multiple nodes, local training is carried out on the basis of a federal learning technology, and the data are obtained. According to the method, deep adaptive learning is combined with the graph neural network, so that the detection precision of an attack mode is improved, particularly modeling of complex attack behaviors is realized, data privacy protection is ensured, and meanwhile, the calculation efficiency is improved.

The invention provides a health data processing method based on distributed privacy protection learning. The health data processing method comprises the following steps: constructing a federal learning framework to generate a machine learning model; performing an iteration process until the model converges or reaches a preset hyper-parameter; the iterative process comprises: independently training the machine learning model by local data; encrypting local data by a preset encryption algorithm, and calculating a gradient to obtain encrypted gradient information; sharing the gradient information based on multi-party computation; combining the gradient information by a preset aggregation algorithm; and updating the parameters of the machine learning model according to the aggregated gradient, and distributing the updated parameters. According to the invention, on the premise of fully protecting user data privacy, efficient and accurate distributed machine learning model training can be realized, so that the overall performance and security of a health data processing system are improved. The invention further provides a health data processing system based on distributed privacy protection learning and the electronic equipment, which have corresponding advantages.

The invention discloses a forest fire monitoring method based on a low earth orbit satellite, and the method comprises the steps: obtaining data through multiple sensors, and carrying out the denoising and smoothing processing, and generating a fusion data set; a convolutional neural network is combined with multi-scale analysis to enhance fire point features, and a feature set is formed. And matching the multi-satellite collaborative observation data with the feature set, and generating a high-frequency observation enhanced data set through time sequence alignment. A deep learning and transfer learning optimization model is adopted, key area features are extracted, multi-satellite data are shared through distributed calculation, and a coverage enhancement data set is obtained. And dynamically adjusting a multi-satellite scheduling strategy based on the timeliness evaluation model, and outputting a scheduling result. The robustness of an algorithm is optimized through reinforcement learning, a robustness enhancement model is obtained in combination with complex meteorological condition simulation, the data fusion precision is verified by using a multi-source data verification technology, and a final monitoring result is output. According to the method, the accuracy, timeliness and coverage integrity of fire point detection are remarkably improved.

The invention relates to a secure transmission method and system for encrypted network data. According to the method, a resource state vector is generated by collecting resource state data of Internet of Things equipment, and an environmental risk level is evaluated by using a pre-training decision model in combination with attacked information; performing content feature scanning on the to-be-transmitted data, and performing data sensitivity grading based on key field identification; querying an encryption decision matrix according to the resource state vector, the environmental risk level and the data sensitivity level, and determining a target encryption scheme; a private key and a public key are dynamically generated at an equipment end, a shared key seed is calculated after the public key is exchanged with a communication opposite end, and a target encryption scheme and the shared key seed are used for encrypting data to generate a transmission frame; periodically optimizing the encryption decision matrix through a reinforcement learning algorithm based on encryption time consumption, resource consumption and transmission success rate indexes; the encryption strength is adaptively adjusted, resource utilization is optimized while the security is ensured, and the overall efficiency of the system is improved through continuous feedback optimization.

The invention relates to an information pushing method and device, computer equipment, a storage medium and a computer program product. The method comprises the steps of obtaining candidate product features and object description features; the candidate product features and the object description features are combined to obtain shared features, at least two autocorrelation degrees of the shared features are calculated, the shared features are transformed based on the at least two autocorrelation degrees, and autocorrelation features corresponding to the at least two autocorrelation degrees are obtained; taking the candidate product features and the object description features as input features, and fusing the input features with the self-correlation features corresponding to the at least two self-correlation degrees in sequence to obtain target fusion features; and calculating an interaction degree of the to-be-pushed object to the product information based on the target fusion feature, and pushing the product information of the candidate product to a terminal of the to-be-pushed object when the interaction degree meets a preset pushing condition. By adopting the method, the information pushing accuracy can be improved.

The invention relates to a data transaction technology, and discloses a privacy enhanced trusted data transaction method based on an alliance chain, which comprises the following steps: a key generation center generates a system parameter list; the data owner calculates a shared key according to the system parameter list, calculates an encrypted ciphertext and uploads the encrypted ciphertext to a decentralized storage network; the decentralized storage network receives the encrypted ciphertext, generates a unique content identifier and divides the encrypted ciphertext into a plurality of ciphertext fragments; calculating a root node hash value, and recording the root node hash value and the unique content identifier into an alliance chain; and the data purchaser encrypts the data demand and issues the data demand through the alliance chain, and the alliance chain matches the transaction data according to the corresponding demand attribute in the data demand and performs decryption operation to obtain the required transaction data. The invention further provides a system of the privacy enhanced trusted data transaction method based on the alliance chain, the secure transaction of the data can be realized, and the controllability of the data in the transaction process is improved.

Neural image compression using substitutional quality factor learning in a latent space, including receiving a compressed bitstream and a target quality factor indicating a target compression quality, calculating a decoded latent representation of the compressed bitstream, and calculating a reconstructed image based on the decoded latent representation of the compressed bitstream and the target quality factor, computing a shared feature based on a network forward computation using shared decoding parameters (SDP) of one or more layers of a convolutional neural network, computing estimated adaptive decoding parameters (ADP) for the one or more layers of the convolutional neural network based on the shared feature, the adaptive decoding parameters, and the target quality factor, and computing an output tensor based on the estimated ADP in the one or more layers of the convolutional neural network and the shared feature.

The invention provides a multi-device communication key negotiation method, and the method comprises the steps: generating a corresponding private key and a scalar multiplication point for each device, the scalar multiplication point being a product of the private key and an elliptic curve base point; each device broadcasts a scalar multiplication point corresponding to the device and receives scalar multiplication points of other devices, and each device calculates a shared key by using a private key of the device and the received scalar multiplication point based on a pairing strategy; encrypted communication between the devices is performed using the shared key. The method has the beneficial effects that the authenticity and integrity of the message are verified by fusing and using certificate signature, the confidentiality of the message is ensured by using the secret key negotiation based on the scalar multiplication point and the private key, the dynamic secret key updating is realized, and the communication security is ensured in multiple aspects.

This application discloses a method, apparatus, device, and medium for item recommendation across multiple user groups, scenarios, and tasks, relating to the field of artificial intelligence technology. The method involves acquiring user-side information, item-side information, and corresponding scene environment information in different scenarios; calculating weight dimensions, shared feature vectors, and environment dimensions to obtain a first dimension, a dense vector, and a second dimension; performing attention mechanism feature cross-calculation to obtain a third dimension; calculating item recommendation scores using the first dimension, dense vector, second dimension, and third dimension to obtain an item recommendation value; determining the target item recommendation value; identifying the target item information corresponding to the item recommendation value; and recommending the target item information to the user. This application can construct a three-dimensional perspective model encompassing multiple scenarios, user groups, and tasks, thereby improving the ability to represent differentiated user groups, meeting diverse user needs, and achieving comprehensive item recommendation across multiple user groups, scenarios, and tasks.

The invention discloses a data sharing system based on the P2P technology, and relates to the technical field of data sharing, data pre-shared by a user is obtained through a client module and is shared to other trusted clients, an equipment scoring unit scores the trusted clients, and the trusted clients are shared by the equipment scoring unit. The sharing analysis unit periodically analyzes the data capacity and click time of all the shared data and calculates a sharing capability evaluation amount in combination with an equipment score, and the sharing management unit intercepts a plurality of trusted clients according to the sharing capability evaluation amount to obtain a sharing management table; then, after a certain client module pre-shares the data, a sharing client is preferentially selected from the sharing management table, and the sharing client replaces the transfer service unit to carry out data sharing operation, so that the load pressure of the server on the shared data can be greatly reduced through the mode; and meanwhile, the shared resources of the trusted client can be reasonably utilized as much as possible.

The invention belongs to the technical field of artificial intelligence, and particularly discloses a deep learning reasoning acceleration method based on calculation sharing. Based on a technical architecture of dynamic computing sharing, a deep learning model is optimized mainly from computing task fine-grained decomposition and common subgraph extraction, construction of computing sharing hierarchical cache driven by shared hidden variables, balanced scheduling of computing sharing loads, a closed-loop prediction feedback mechanism of a computing sharing strategy and the like; according to the method, the reuse capability of the calculation sub-graph and operator level and the cache hit rate of forward reasoning weight are improved, and dynamic sub-graph division and migration are supported, so that the speed-up ratio and the resource utilization rate of a typical model are improved. And in combination with parallel computing, dynamic resource scheduling and computing resource sharing mechanisms, the reasoning efficiency in the heterogeneous computing environment is optimized.

The invention discloses a multi-source data collection and cross-system calling method based on cloud edge collaboration, and belongs to the technical field of cross-system interaction. According to the method, edge nodes and business software interfaces are uniformly coded to complete collection, and then interface interaction relation pairs are constructed and a data interaction behavior set is formed; an interaction timestamp is configured through interface heartbeat, and the data transmission log cluster is mapped into a signal map; calculating a shared cooperation gap to expand and form a complete interaction link, and further constructing a link state matrix; and finally, evaluating the data calling sharing degree between the edge nodes based on the matrix, and screening replaceable nodes when the data leakage risk occurs in the nodes. According to the method, through standardized collection, mapping analysis, precise link construction and sharing degree evaluation, the problems that in the prior art, data collection is disordered, cross-system calling is unstable, and fault nodes are difficult to replace are solved, efficient integration and cross-system smooth calling of multi-source data are achieved, and the reliability and fault tolerance of the system are improved.

The application discloses a logistics cost management method and system based on shared storage resources, comprising the following steps: step one: collecting original actual operation data of each warehouse in the shared storage platform, and performing time alignment and correction processing; step two: establishing a shared storage resource network diagram; step three: obtaining multiple complete cost occurrence paths, calculating various logistics costs, and obtaining a logistics cost set; step four: inputting the shared storage resource network diagram and the logistics cost set into an improved Graphormer model, obtaining a storage cost time sequence attention result through a resource occupation coding module, a cost connection enhancement module and a storage time sequence attention module; and step five: extracting a high-cost conduction risk path according to the storage cost time sequence attention result, calculating a shared warehouse comprehensive cost, and obtaining a shared storage resource reallocation scheme. The improved Graphormer model is used to realize intelligent reallocation of shared storage.

The application relates to data transaction technology and discloses a privacy-enhanced trusted data transaction method based on a consortium chain, which comprises the following steps: a key generation center generates a system parameter list; a data owner calculates a shared key according to the system parameter list, calculates encrypted ciphertext, and uploads the encrypted ciphertext to a decentralized storage network; the decentralized storage network receives the encrypted ciphertext, generates a unique content identifier, and divides the encrypted ciphertext into multiple ciphertext fragments; a root node hash value is calculated, and the root node hash value and the unique content identifier are recorded in the consortium chain; a data buyer encrypts data demand, publishes the data demand through the consortium chain, and the consortium chain matches transaction data according to corresponding demand attributes in the data demand and performs decryption operation to obtain required transaction data. The application further provides a system of the privacy-enhanced trusted data transaction method based on the consortium chain, which can realize safe transaction of data and improve controllability of data in a transaction process.

The embodiments of the present disclosure relate to a method, device and medium for realizing data arrangement, the method comprising: based on the number of threads of a thread group and data arrangement requirements, obtaining first auxiliary data and second auxiliary data; shifting data in a first register, the first auxiliary data, the data in a second register and the second auxiliary data to a transfer register; based on the address of the transfer register, extracting the data in the transfer register, thereby obtaining first data, second data, third data and fourth data; based on the number of threads of the thread group and data arrangement requirements, determining mapping parameters; and based on the mapping parameters, mapping the third data to the first register and mapping the fourth data to the second register, thereby realizing data arrangement in registers. Thus, data arrangement between multiple registers can be realized more quickly without calculating a shared memory address.

To provide a calculation server and a communication terminal that communicates with the calculation server, which calculate the shared ride distance and the proportion of the usage fee borne by users based on the distance traveled, without increasing the burden on users and drivers of paid vehicles. [Solution] The processing method of an information processing device that communicates with a first terminal of a first user and a second terminal of a second user riding in a vehicle involves transmitting information regarding the first fare to be borne by the first user to the first terminal and information regarding the second fare to be borne by the second user to the second terminal, based on first boarding location information based on wireless communication between the vehicle's electronic device and the first terminal, first disembarking location information based on the interruption of wireless communication between the electronic device and the first terminal, second boarding location information based on wireless communication between the electronic device and the second terminal, and second disembarking location information based on the interruption of wireless communication between the electronic device and the second terminal.

The invention discloses a disaster recovery redundant link optimization method and device for a smart power grid dispatching communication network, and the method comprises the steps: firstly obtaining the topology and link attributes of the dispatching communication network, mapping the time delay, rental or bandwidth and other indexes into weights, calculating a shortest main path from a source dispatching center to a target substation on an original graph, and carrying out the calculation of the shortest main path from the source dispatching center to the target substation; constructing a split residual image based on the path, and reserving a cost bit of a shared link in the split residual image; a candidate standby path with the shared link number not exceeding a threshold value is obtained through a limited shortest path algorithm, a final redundant path pair is generated in combination with the main path, the total weight of the final redundant path pair is calculated according to a union set, and the weight of the shared link is metered only once; and finally, issuing the obtained main and standby paths to electric power communication network equipment to realize rapid switching when the link fails. According to the method, the construction and operation and maintenance cost is reduced while the reliability is ensured, the resource utilization rate is improved, and good deployment feasibility and near-linear time complexity performance are achieved.

To provide a technique for an appropriate use charge corresponding to a disturbance degree and collection easiness of returned shared mobility.SOLUTION: A shared mobility use charge calculation method acquires return position information showing a return position of a shared mobility rented to a user, calculates disturbance degree of a returned shared mobility or collection easiness of the shared mobility on the basis of the return position information, and calculates a use charge of the shared mobility on the basis of the disturbance degree or the collection easiness.SELECTED DRAWING: Figure 1

The present invention discloses an encryption communication method and system for metadata business services, the method comprising: a first business application party sends a business metadata call request to a second business application party, the second business application party generates a second random number according to the business metadata call request, and the second business application party generates a second intermediate parameter according to the second random number; the first business application party generates a first random number by itself, wherein the first business application party generates a first intermediate parameter according to the first random number; the first business application party sends the first intermediate parameter to the second business application party, and the second business application party sends the second intermediate parameter to the first business application party; the first business application party and the second business application party respectively calculate a shared key according to the second intermediate parameter and the first intermediate parameter, and determine the business metadata call authorization according to the shared key.

Various embodiments of a method and apparatus are disclosed for creating a new device that implements an algorithm, subject to specified constraints. In some embodiments, an initial algorithm and constraints are received and converted to a new algorithm, upon which the device is based. The method further includes constructing a DAG (Directed Acyclic Graph) from the algorithm received and then reconstructing the DAG to accommodate the constraints. The method and system identify outputs that are of interest, trace the outputs of interest back to the inputs, and ignore inputs and the parts of the DAG that are not needed for generating the outputs of interest. When computing multiple jobs in parallel that each use the same DAG, portions of inputs that are shared by two parallel jobs, and portions of the DAG that compute the shared inputs are determined, and that therefore only need to be computed once.

Provided is a technology for performing secure computation of a random number generation method using weighted probability distribution with high accuracy while keeping data secure. The technology includes: first vector computation means that computes a share ([[p′1]], . . . , [[p′L]]) from a share ([[p1]], . . . , [[pL]]) by using prefix sum; uniform random number generation means that generates a share ([[q1]], . . . , [[qS]]) of a vector (q1, . . . , qS) (where qi (i=1, . . . , S) is a uniform random number, and satisfies 0≤qi≤1) having a uniform random number as an element; and random number computation means that computes a share ([[r1]], . . . , [[rS]]) of a vector (r1, . . . , rS) having an output value as an element from the share ([[p′1]], . . . , [[p′L]]), a share ([[x1]], . . . , [[xL]]), and the share ([[q1]], . . . , [[qS]]) by using secure collective mapping.

The application discloses a network user classification method based on adaptive clustering center clustering. The method is based on the idea of DPC clustering algorithm, and attribute division is performed according to user behavior data features. According to the distance between data points and other data points, the combined density value of each user behavior is calculated, and noise points, boundary points and normal points are divided. The noise points are discarded, the normal points are first divided into sub-clusters, then the representative points of the sub-clusters are subjected to distance transformation by calculating shared neighbors, the final clustering center, i.e. the user attribute label, is selected from the representative points of the sub-clusters. Finally, according to the similarity between the sub-clusters, the boundary points are divided into the category in which the nearest normal point is located. The method has few required parameters, strong adaptive capacity and high automation degree of the clustering process, can automatically determine the classification category according to the data distribution characteristics, does not need manual determination of attribute features or quantity, and can effectively distinguish users with different attributes.

Methods for distributing shared secret keys in communication networks using network coding techniques that enable terminal nodes to compute keys directly without reconstructing source information. Source nodes generate random bits that are transmitted through the network via intermediate nodes performing network coding operations. Terminal nodes compute shared secret keys as functions of the received coded messages without the need for recovering individual source bits. The methods provide security against eavesdroppers at individual nodes or specified edge sets while achieving higher key rates than traditional approaches requiring source reconstruction. Multiple terminal sets can obtain pairwise independent keys through appropriate labeling schemes. Embodiments employs padding protocols to protect information through cut vertices and supports both single and multiple source configurations.

The invention belongs to the field of data security, particularly relates to a data sharing method and system and a computer equipment cluster, and aims to solve the problems of selecting expected data and verifying whether the data is tampered or not. The method comprises the following steps: a data owner encrypts shared data by using a first private key, generates a summary description of the shared data, and calculates a first hash value of the shared data; how to store a cloud server storage address, a summary description and a hash value of a ciphertext to a block chain; after a data use application of a data user is received, a re-encryption key is generated according to the first private key and a second public key of the data user, the proxy encrypts a ciphertext for the second time according to the re-encryption key and sends the ciphertext to the data user, the data user decrypts the ciphertext subjected to the second time encryption to obtain shared data, a second hash value is generated according to the shared data, and the second hash value is sent to the data user; and verifying whether the second hash value is consistent with the first hash value on the block chain. According to the method, multi-stage credible management is adopted, and privacy, safety and integrity of shared data are guaranteed.

The application discloses a kind of forest fire monitoring methods based on low-orbit satellite, comprising: obtaining data by multiple sensors, and generating fusion dataset by denoising and smoothing processing.Utilize convolutional neural network to combine multi-scale analysis to enhance fire point features, form feature set.Multiple satellite cooperative observation data are matched with feature set, and enhanced dataset of high frequency observation is generated by time series alignment.Adopt deep learning and transfer learning optimization model, extract key area features, share multiple satellite data through distributed computing, obtain coverage enhanced dataset.Based on timeliness evaluation model, dynamically adjust multiple satellite scheduling strategy, output scheduling result.Through reinforcement learning optimization algorithm robustness, combined with complex weather condition simulation, obtain robustness enhanced model, and utilize multi-source data verification technology to verify data fusion precision, output final monitoring result.The application significantly improves the accuracy, timeliness and coverage integrity of fire point detection.

The present invention discloses an intelligent power grid identity authentication system based on physical unclonable functions, which includes: a power grid intelligent device sends Msg1 containing challenge information to a gateway; the gateway generates Msg2 based on Msg1 and sends it to a control center; the control center confirms the identity of the power grid intelligent device based on the challenge information in Msg2, authenticates the gateway and the power grid intelligent device based on Msg2 and calculates a shared key SK; the gateway authenticates the control center based on Msg3 and calculates the shared key SK, and sends Msg4 to the power grid intelligent device; the power grid intelligent device calculates the shared key based on Msg4, re-responds to the challenge information based on its physical unclonable function, and authenticates the control center using the challenge response information. The present invention realizes effective identity authentication in the intelligent power grid system, can resist various network attacks and physical attacks, and the operations are lightweight.

The invention relates to the technical field of privacy protection technologies, in particular to a federated learning method for resisting a lagging-behind person based on summer secret sharing, and aims to solve the problem of federated learning of the lagging-behind person caused by different client resources and different computing performances by adopting an improved summer secret sharing scheme. In the initialization stage of training, user data and user model parameters are distributed to different users by using summer secret sharing, so that the training stage is a data training stage of all the users, and embedded aggregation of all the clients is rebuilt losslessly by decrypting calculation sharing of non-serial clients; and moreover, a summer secret sharing scheme is used for data distribution and encryption, so that the security of the data in the transmission process is also ensured. According to the method, a homomorphic encryption scheme is replaced with a summer secret sharing scheme, complex ciphertext calculation of homomorphic encryption is replaced, and the complexity of data interaction calculation in federated learning is reduced.

The invention relates to the technical field of robot control, in particular to a force sense control method fusing robot visual information, and solves the technical problems of operator intention distortion and man-machine control right conflict. The method comprises the following steps: identifying an interaction object through scene semantic segmentation, and establishing a virtual force model to generate a predictive guide data stream; receiving an operating force of an operator through a force sensor, executing online dynamic compensation to obtain a pure intention force, and analyzing the pure intention force by applying an algorithm to obtain an operating instruction vector; applying a dynamic projection fusion algorithm, taking the modulus length of the guide data stream as a dynamic weight, performing non-conflict fusion on the guide data stream and the manipulation instruction vector, and calculating a shared motion track; and finally, outputting a trajectory tracking signal by adopting a position tracking controller, and providing force sense guidance for an operator through the active movement of the robot. According to the invention, efficient, safe and smooth man-machine cooperative control is realized.

The invention provides a secret key dynamic distribution method and system for an intelligent network connection vehicle, and belongs to the field of vehicle communication information security and technology. The key dynamic distribution method comprises the following steps: setting and storing key information; the vehicle end and the cloud server end establish a secure communication channel, the vehicle cloud establishes a communication session, the cloud server end calculates a key strength demand value, obtains a key strength demand level and calculates a shared key value, and the vehicle cloud negotiates a shared key and uses the shared key to perform data secure transmission; and when the communication session ends or the data communication time threshold value is reached or the data communication traffic is reached, the key is replaced. The key dynamic distribution system comprises a cloud server side and a vehicle side, the cloud server side is provided with a cloud key public opinion module, a password information base, a cloud key management module, a cloud service management module and a cloud communication management module, and the vehicle side is provided with a communication management module, a service management module and a key management module. According to the invention, the security of data communication is improved, and the risk of data leakage is reduced.