91results about How to "Guarantee privacy and security" patented technology

The invention discloses a hierarchical multi-agent game confrontation and collaborative decision-making algorithm based on federated learning, and the algorithm comprises the following steps: setting a federated learning framework containing a plurality of agents, enabling the agents to be located in different regions, and carrying out the training of a local model through the data of the regions; in the federated learning framework, deploying a layered reinforcement learning model in each intelligent body; performing confrontation training on a plurality of agents in a game environment by using a hierarchical reinforcement learning model; on the basis of game confrontation, a collaborative decision algorithm is introduced to coordinate behaviors of multiple agents; and when a preset training round is reached or a specific condition is satisfied, uploading local model parameters to a central server, and sharing the model parameters through a federal learning framework to carry out global optimization. According to the method, efficient collaboration and strategy optimization among multiple agents can be realized on the premise that privacy is protected, and high adaptability and robustness are kept in a complex and changeable environment.

The invention provides a security management and retrieval method based on Internet of Vehicles data, and the method comprises the steps: carrying out the cooperative training of a global key management model through a federated learning technology on the premise of not leaking the original data of all parties, and achieving a security key management scheme under the participation of multiple parties through a model parameter aggregation and updating mechanism; when data retrieval is needed, a user submits a retrieval request and identity authentication information, and after a system verifies the identity of the user, a ciphertext data identifier meeting conditions is matched in an attribute index according to a keyword and an authorization attribute of the request; the system acquires corresponding ciphertext data from a distributed database according to the ciphertext data identifier, and directly performs keyword matching calculation on the ciphertext data through a secure multi-party computing protocol under the condition of not decrypting the data; and according to the mined security threats and abnormal modes, a data encryption strategy and an access control strategy are dynamically adjusted, authority limitation or access blocking is performed on suspicious users, and the security protection capability of the system is improved.

The invention discloses a diagnostic data remote transmission method based on a regional remote diagnosis and treatment platform. The method comprises the following steps: S1, collecting diagnostic data of a patient; s2, preprocessing the acquired diagnosis data; s3, carrying out data transmission by utilizing a short-distance wireless communication technology and a long-distance mobile communication technology; s4, performing data encryption and identity verification by adopting AES and RSA encryption algorithms; s5, after the regional remote diagnosis and treatment platform receives the diagnosis data, decryption and identity verification are carried out on the data; s6, performing image data analysis by using a convolutional neural network, processing character data by using a natural language processing technology, predicting and classifying physiological parameters by using regression analysis and a classification algorithm, and performing fusion analysis on multi-modal data in combination with a deep learning model; and S7, transmitting the analyzed and processed diagnosis result back to a local data terminal through a long-distance mobile communication technology. The system has the advantages of high transmission stability, high safety, high transmission efficiency and high diagnosis accuracy.

The embodiment of the invention provides a snore recognition method and system based on a non-contact sensor, and the method comprises the steps: obtaining vibration data collected by the non-contact sensor, and carrying out the preprocessing of the vibration data, and obtaining vibration signals of a plurality of time windows; calculating a time domain feature, a frequency domain feature and a time frequency feature of the vibration signal; training a weak classifier model, and training a strong classifier by taking a prediction result of a weak classifier as an input feature; collecting a real-time vibration signal, inputting the time domain feature, the frequency domain feature and the time frequency feature of the real-time vibration signal into the weak classifier to obtain an output prediction probability value, and inputting the prediction probability value into the strong classifier to obtain an output snore recognition result; and detecting a snore recognition result, correcting the snore recognition result based on a detection result in combination with the prediction probability value, and outputting a final snore result after correction.

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 RAG data acquisition and question and answer generation method and system based on an encyclopedia website, and relates to the technical field of data processing.The method comprises the steps that entry content in the encyclopedia website is captured, encrypted and stored in a database, and an index is established; extracting semantics of the vocabulary entry content to generate a knowledge graph; obtaining interaction data of the user, and carrying out interaction willingness analysis to identify interaction willingness data of the user for each entry; dynamically updating the database; and generating content reply according to the knowledge graph. The method comprises the following steps: capturing entry contents of an encyclopedia website in real time, combining data cleaning, encryption and index construction technologies to ensure the security and efficient and timely retrieval of data, constructing a knowledge graph based on semantic information of entries, realizing information structuring and semantic association, and generating personalized reply of a user based on a knowledge graph retrieval and generation technology. According to the overall scheme, the data real-time performance, the content generation accuracy and the user experience are remarkably improved, and meanwhile data privacy and system safety are guaranteed.

The invention provides a load balancing routing method and system of a satellite network. The method comprises the following steps: performing satellite operation state acquisition on satellite nodes through a satellite sensor network to obtain satellite original state data; performing network state sensing on the satellite original state data to obtain a satellite network state data set; performing task demand mining on the communication participation nodes to obtain a task demand trend data set; and performing task demand prediction modeling and task demand prediction according to a preset deep learning model and the task demand trend data set to obtain a task demand prediction data set. Through the technical methods of satellite state perception, task demand prediction, distributed resource optimization, reinforcement learning resource scheduling, a privacy protection mechanism and the like, the energy efficiency of a multi-satellite system is improved, task load balance and coordination in a cloud edge collaborative environment are realized, and privacy security in an air-space-ground integrated network is guaranteed.

The invention provides a fatigue driving adaptive detection method, which comprises the following steps: step S101, acquiring facial features of a driver during driving, and calculating a fatigue index of the driver according to the facial features of the driver and a fatigue monitoring calculation method; s102, when the calculated fatigue index exceeds a set fatigue index normal threshold value, a fatigue driving signal is sent to the vehicle; s103, if the driver does not have the fatigue driving behavior, re-judging the type to which the eyes of the driver belong; step S104, modifying parameters in a fatigue monitoring calculation method according to the type of the eyes of the driver, and performing adaptive adjustment on the fatigue monitoring calculation method; and S105, circulating the above steps by using the adjusted fatigue monitoring calculation method until the monitoring is normal. According to the invention, the problem of false alarm of the fatigue monitoring system caused by small eyes of the driver is solved, and the accuracy and adaptability of fatigue monitoring are improved.

The invention discloses a voice-driven virtual scene generation and switching method and an implementation system, which are applied to meta universe VR (virtual reality) glasses equipment, and comprise the steps of recognizing an instruction of a user through voice recognition and natural language processing technologies, generating or switching a virtual scene in real time, and realizing personalized scene creation and object generation. The edge computing node is adopted to process a high-calculation-amount task, the delay of scene generation and switching is remarkably reduced, the data transmission efficiency is improved by dynamically selecting the optimal node, the privacy security of a user is guaranteed, and through frame rate smooth adjustment and a scene fade-out mechanism, the experience of the user for exiting the virtual scene is optimized, the dizziness is reduced, and the user experience is improved. And the interaction fluency and comfort are improved. The method has the characteristics of quick response, convenient operation and strong privacy protection, and is widely applied to the virtual reality application fields of games, education, medical treatment and the like.

The invention relates to a light leakage prevention diffraction optical waveguide which comprises a waveguide substrate, an input diffraction optical element and an output diffraction optical element. Incident light is coupled into the waveguide substrate through the input diffractive optical element, and is coupled out of the waveguide substrate step by step through the output diffractive optical element after being totally reflected in the waveguide substrate; the input diffractive optical element and the output diffractive optical element are respectively and independently arranged to be of a periodic grating structure, and a grating of the output diffractive optical element is arranged to be an asymmetric surface relief grating. The grating tilt direction of the output diffractive optical element and at least one further grating structure parameter are modulated such that the diffraction efficiency of the output diffractive optical element towards the out-coupling side exceeds the diffraction efficiency away from the out-coupling side, and the diffraction efficiency towards the out-coupling side varies spatially. According to the invention, the inclined grating or blazed grating easy to process is used in the coupling-out area, and the optical performance of the diffraction optical waveguide is considered by modulating the grating parameters, so that the effects of effectively weakening light leakage and ensuring the privacy of a user are achieved.

The application discloses a screen projection method, device and system, electronic equipment and a storage medium. The screen projection method comprises the following steps: receiving a broadcast signal; analyzing the broadcast signal to obtain screen projection information; judging whether to display a screen projection request prompt based on the screen projection information; if it is judged to display the screen projection request prompt, the screen projection request prompt is displayed; and in response to a confirmation operation of the user agreeing to the screen projection request prompt, a screen projection operation is performed. The application triggers the screen projection operation by the screen projection end, specifically, the broadcast signal sent by the screen projection end is received, and the screen projection operation is performed according to the screen projection information carried in the broadcast signal, so that the user does not need to search for the screen projection end on the screen projection end, the operation is convenient, the error rate of the screen projection connection is reduced, and the user experience is improved.

The application discloses a kind of emotion evaluation and intervention method and system based on multi-modal data.Through terminal synchronous acquisition voice, face, screen tremor and so on multi-modal data;Reaction mode vector is generated based on context information and data verification fusion;The composite emotional state of user is determined using emotional emergence model.At the same time, personalized emotional causal diagram is constructed and updated, and the enhanced state vector is generated by extracting key causal path features.The vector is input into the intervention agent to output intervention actions;After execution, the emotional state is evaluated again and the reward is calculated, and the system model is updated online.The application realizes accurate emotion evaluation and closed-loop adaptive intervention that strikes the root.

The present application relates to the technical field of vehicle pseudonym replacement, and particularly relates to a vehicle pseudonym replacement method based on artificial intelligence, which comprises the following steps: determining whether pseudonym replacement is needed according to a replacement time interval value or performing secondary determination based on a quantity ratio reference value and a communication characteristic value; determining a target area type according to a to-be-replaced vehicle distribution reference value and a replacement allowance, and determining whether to directly perform pseudonym replacement or perform combined analysis according to the target area type; in the combined analysis, determining a related vehicle combination based on a position synchronization degree or an interaction reference value according to a vehicle analysis coefficient, determining whether to perform sequential setting of a virtual vehicle based on a quantity difference value or setting a virtual vehicle based on a historical attack frequency of a sub-area, and performing pseudonym replacement on to-be-replaced vehicles and virtual vehicles in the related vehicle combination. The present application can reduce the risk of vehicle location privacy leakage.

The invention relates to the technical field of traditional Chinese medicine intelligent pulse feeling, and discloses a wearable traditional Chinese medicine intelligent pulse feeling watch and a pulse feeling method.The watch comprises a watch body, a first wrist strap and a second wrist strap, the first wrist strap and the second wrist strap are fixedly connected to the left side and the right side of the watch body respectively, and a clamping groove is formed in the middle of the first wrist strap; the outer side of the second wrist strap is fixedly connected with a clamping block, the clamping block is matched with the clamping groove in an inserted mode, an air bag is installed in the first wrist strap, a notch is formed in the middle of the second wrist strap, and the first wrist strap is connected to the middle of the notch in a sliding mode. Through the high-precision sensing array and the signal processing module, pulse signals are accurately collected, motion interference is eliminated, the measurement stability and accuracy are improved, meanwhile, the data encryption and cloud synchronization technology is adopted, user privacy safety is ensured, long-term tracking and remote management of health data are achieved, and the user experience is improved. The problems of signal instability, motion interference and data security in the prior art are solved.

The application discloses a kind of stealthy photographing behavior detection method, device, electronic equipment and storage medium, the present application first collects the pedestrian image in target area, then, through machine recognition technology, to identify the arm posture of each pedestrian in pedestrian image, and determine the stealthy photographing suspect object according to arm posture;Then, again through the hand-held object type of identifying the stealthy photographing suspect object, and the dress type of pedestrian in its corresponding stealthy photographing image area, to finally determine whether the stealthy photographing suspect object exists stealthy photographing behavior;Finally, after determining that there is stealthy photographing behavior, then the interception of stealthy photographing image can be carried out, and sent to monitoring terminal, to realize the retention of evidence and secondary detection of stealthy photographing behavior;Therefore, the present application uses machine recognition technology to realize stealthy photographing behavior detection, compared with artificial patrol, efficiency is higher, detection accuracy is higher, effectiveness is stronger, can produce greater deterrent to lawbreaker, can fully guarantee the privacy security of passenger.

This invention belongs to the field of identity authentication in the Internet of Vehicles (IoV), specifically relating to an anonymous batch authentication method for IoV commuting environments. The invention includes: constructing an IoV system comprising a Trusted Authority (TA), roadside units (Roadside Units), and vehicles; publishing and initializing system parameters; vehicles and Roadside Units pre-registering and verifying their identities with the TA; the TA verifies the identity, calculates and returns key parameters and a pseudonym; vehicles calculate their complete private and public keys based on the returned parameters and send a verification message to the TA; the TA verifies the identity and creates a vehicle authentication schedule; commuting vehicles initiate authentication to Roadside Units (Roadside Units) on the commuting route; the Roadside Units receive the message broadcast by the vehicles, verify the legitimacy of the vehicle requests using the vehicle authentication schedule, and complete the authentication. This invention ensures anonymous and secure communication among entities in the IoV system and reduces computational and communication overhead, thereby improving authentication efficiency.

The invention provides a password device discovery method and a computer device, and relates to the technical field of password application. The method comprises the following steps: responding to an access event triggered by a software application on a current password device, obtaining device feature information of the current password device in multiple dimensions, generating device fingerprint information of the current password device according to the device feature information in multiple dimensions, and sending an access request to a server of the software application according to the device fingerprint information, the access request is used for enabling the server to detect the equipment fingerprint information by adopting an equipment fingerprint detection algorithm according to a preset equipment fingerprint database to obtain a fingerprint detection result of the current password equipment, receiving a first access response sent by the server, and sending the first access response to the server; the first access response is an access response triggered when the server determines that the current password device is the registered password device. Complicated and unique equipment fingerprint information is generated through multi-dimensional equipment feature information, and omnibearing and high-precision identification of the identity of the password equipment is realized.

The application relates to a user data storage management method and an embedded device. The method is applied to the embedded device, user conversation data is acquired, local end information extraction is performed on the user conversation data based on preset extraction conditions, current user key data is obtained, original user key data stored in a first storage unit is acquired according to the current user key data, hierarchical storage judgment is performed on the current user key data and the original user key data, first storage data meeting first preset storage conditions is stored in the first storage unit, second storage data meeting second preset storage conditions is stored in a second storage unit, a user search instruction is acquired, the first storage unit and the second storage unit are searched according to the user search instruction and a preset search algorithm, target data is obtained, user privacy data leakage is avoided, user privacy safety is guaranteed from an architecture level, and the reliability of storage management in a resource-limited platform is improved.

The application relates to the technical field of artificial intelligence, and discloses a large model expert library construction based on a knowledge graph, which realizes real-time acquisition of multi-source data, marks an update point through a hash fingerprint algorithm and semantic similarity calculation, and stores the update point in a message queue in combination with metadata determination; a low-dimensional semantic vector is generated by extracting features and constructing a fusion network, and cross-modal mapping is established; an attribute graph is built with the semantic vector as a node, and the attribute graph contains basic facts, reasoning rules and time sequence edges; a user request is parsed as a graph traversal task, and a result is returned through counterfactual reasoning; sensitive data and relationships are encrypted and indexed; source field graph features are migrated to optimize target field graph, and the construction of the expert library is accelerated. The application realizes efficient data processing, semantic correlation analysis, safe query and rapid construction of the expert library, and improves data utilization and knowledge reasoning efficiency.

The invention relates to the field of federated learning, in particular to an aircraft manufacturing equipment abnormity diagnosis method based on federated learning. According to the method, firstly, a federal learning training framework is adopted for training of an aircraft manufacturing equipment anomaly detection model, the privacy security of data is guaranteed, and the privacy security of the aircraft manufacturing equipment data is further enhanced. On the basis of guaranteeing the data privacy security, the model for detecting the abnormality of the aircraft manufacturing equipment is trained, and model training is carried out on the data of the aircraft manufacturing equipment under the condition that the data are not out of local equipment through a federal learning architecture, so that the privacy security of the data is guaranteed. According to the method, a contract theory incentive mechanism is adopted, reasonable reward is provided for the aircraft manufacturing equipment to participate in the federated learning training process, the aircraft manufacturing equipment is incentive to participate in the federated learning training, and the accuracy of the aircraft manufacturing equipment anomaly detection model is improved.

The invention discloses a user side self-discipline voltage regulation method and device considering responsibility definition, and the method comprises the steps: S1, obtaining a unit impedance value and length of a power distribution line by a power grid company, and carrying out the calculation to obtain the total resistance and reactance of the line from a user node to a power supply node; s2, the power grid company sets the reference active power and reactive power of the user according to the power consumption condition of the user and issues the set values to the user side; s3, the user obtains active power and reactive power of the user; s4, the user calculates an active power fluctuation quantity and a reactive power fluctuation quantity of the user; s5, the user calculates an active power fluctuation ratio and a reactive power fluctuation ratio of the user; s6, the user judges whether at least one of the own # imgabs0 # exceeds a reference threshold value or not, if yes, a success rate out-of-limit signal Sig1 is generated, and the step S7 is executed; if not, generating a power non-out-of-limit signal Sig2 and executing the step S3; s7, the user calculates the self-discipline compensation amount required by the user; and S8, the user uses a parallel capacitor, an SVG static var generator and other devices to carry out on-site compensation. According to the invention, the problem that the voltage regulation responsibility sharing is not clear is solved, and the privacy security of the user is also ensured.

The invention discloses a service-oriented robot behavior recognition method based on a PAF vector field algorithm and a depth map, and relates to the technical field of artificial intelligence and robots, and the method comprises the steps: employing a depth camera to obtain a depth image; human body information in the foreground of the depth image is separated from the background by adopting a foreground and background separation algorithm, and a background image is output; 3, processing the depth image by using a PAF vector field algorithm, extracting skeleton key points of a human body, generating skeleton information, and obtaining a human body posture according to the skeleton information; 4, calling the background image by adopting a privacy protection module, and placing the skeleton information in the background image; and 5, detecting whether a falling event or an abnormal waving action exists or not by analyzing the extracted human body posture, and if so, triggering an alarm mechanism. The human body posture in the service area is detected in real time through the PAF vector field and the depth camera, high real-time performance and high accuracy are achieved, and falling detection requirements in various complex environments can be effectively met.

The present invention provides a load balancing routing method and system for a satellite network. The method comprises the following steps: collecting satellite operation status of satellite nodes through a satellite sensor network to obtain satellite original status data; performing network status perception on the satellite original status data to obtain a satellite network status data set; performing task demand mining on communication participating nodes to obtain a task demand trend data set; performing task demand prediction modeling and task demand prediction according to a preset deep learning model and a task demand trend data set to obtain a task demand prediction data set. The present invention improves the energy efficiency of a multi-satellite system, realizes task load balancing and coordination in a cloud-edge collaborative environment, and ensures privacy security in an air-ground integrated network through technical methods such as satellite status perception, task demand prediction, distributed resource optimization, reinforcement learning resource scheduling, and privacy protection mechanisms.

The invention provides a multi-layer illusion special effect resource generation method and device, equipment and a storage medium, and relates to the technical field of image processing. The method comprises the following steps: displaying a phantom multi-layer configuration interface; receiving a resource packet corresponding to each layer; carrying out blank area identification on the sequence frame image and the default static image of each image layer, calculating a uniform cuttable area, and carrying out cutting processing to obtain a target sequence frame image and a target static image of each image layer; based on a dragging operation of a user or an offset obtained by form input, adjusting the offset of the offset layer relative to the fixed layer, and outputting a target offset; and converting the target sequence frame images of each image layer into target animations in batches, exporting the target animations, generating a JSON configuration file, compressing and packaging the target animations, the target static images and the JSON configuration file, and generating a illusion special effect resource packet. And independent management of multi-layer resources, automatic identification and cutting of blank areas and accurate offset control of the layers are realized.

The invention provides an AI video monitoring platform based on an RISC-V framework. The AI video monitoring platform comprises eight levels of frameworks including an equipment perception layer, an edge computing layer, a data access layer, a data processing layer, a data storage layer, an algorithm engine layer, a platform application layer and an industry application layer. The invention belongs to the technical field of monitoring platforms, particularly relates to an AI video monitoring platform of an RISC-V architecture, and aims to at least solve the problems that traditional security depends on manual screen staring and inspection, abnormal event recognition lags behind and the missing report rate is high, and the existing system is high in authorization cost, insufficient in edge scene energy efficiency ratio and poor in adaptability. The risks that the data transmission delay is high, real-time early warning cannot be supported and privacy leakage exists in original video cloud storage due to a front end-cloud framework are avoided; the method avoids the phenomena that different brand devices and platform protocols are not uniform, the compatibility is poor, the deployment and maintenance cost is high, and the expansion capability is weak, and overcomes the defects that a general algorithm is insufficient in precision, lacks an industry exclusive algorithm, and cannot cover the multi-scene safety protection requirement.

The present invention discloses a resource synchronization coordination method and system based on fully homomorphic encryption. First, the resource information of any two users is homomorphically encrypted to obtain an encrypted ciphertext; then the larger of the ciphertexts is assigned a value of m, and the smaller of the ciphertexts is assigned a value of n; and the greatest common factor of the encrypted ciphertexts is determined, and the least common multiple is determined according to the ciphertext n and the greatest common factor; and the resources are coordinated and synchronized using the least common multiple. The application can realize real-time, efficient and secure synchronization and coordination of resources without leaking user resource information.

The invention relates to a smart home and smart property agent cooperation system and method, the system comprises a home agent, a property agent and a community agent, and the home agent is used for processing multi-modal input of a user according to a personalized decision model and generating event information according to a privacy strategy. And the property agent is used for performing task reasoning according to the event information and a preset knowledge base and generating a corresponding task instruction. And the community agent is used for executing the task instruction and feeding back an execution result to the property agent. Wherein the home agent, the property agent and the community agent perform data interaction through a preset communication architecture. The problem of how to improve the business processing efficiency while the intelligent agents of the intelligent home and the intelligent property management system cooperate in the related technology is solved.

The invention discloses a multi-person shared health detection device capable of accurately recognizing user identity based on sound. The device comprises a microphone array, a physiological parameter sensor, a processor and a memory. The method comprises the following steps of: performing front-end processing on a collected voice signal to obtain an effective voice segment; extracting a voiceprint embedding vector by using a deep neural network; comparing the vector with a local voiceprint template library to obtain a highest score and candidate identities; carrying out living body and playback attack authenticity discrimination by analyzing multi-dimensional acoustic features and utilizing a classification model; and binding and associating the identity label with the health data collected by the sensor when the identity label is judged to be a real living body and the similarity score meets a judgment condition. The method has the advantages of high noise resistance, attack resistance and end-side low delay, supports offline work and model increment self-adaption, is suitable for families, communities, medical institutions and other scenes, and effectively improves the use convenience and data security of shared health equipment.

The application relates to the field of video applications, and provides a video desensitization method and device, electronic equipment and a computer program product. The method comprises the following steps: acquiring video data, identifying a human hand region based on the video data, obtaining N gesture recognition results and M hand contact article recognition results, determining a sensitive region and a target sensitive level according to the N gesture recognition results and the M hand contact article recognition results, and performing desensitization processing on the video data according to the sensitive region and the target sensitive level. The video desensitization method provided in the embodiment of the application can solve the technical problem that a hand contact article or hand operation of a photographed person cannot be desensitized, personal information of the photographed person is easily leaked, and the personal information safety of the photographed person is improved.