59results about How to "Ensure fairness" patented technology

The invention discloses an LLM-based graduation design thesis intelligent evaluation system. The system comprises a data input module, a preprocessing module, an LLM evaluation module and an evaluation result output module. Text data of graduation papers are input into the data input module, the data input module is connected with the preprocessing module, the text data, processed by the preprocessing module, of the graduation papers are input into the LLM evaluation module, and the evaluation result output module is connected with the LLM evaluation module. According to the system, a large language model is utilized to perform multi-dimensional evaluation on graduation theses submitted by students, including but not limited to content quality, logic structure, language expression, innovation, academic specification and the like. According to the LLM-based undergraduate thesis evaluation system, the LLM can adapt to thesis evaluation requirements in different subject fields through a fine adjustment mode.

The invention relates to a class increment image classification method and system based on knowledge distillation. The method comprises the following steps: generating output representations of old class training samples among old classes through a teacher model, and respectively carrying out Z score standardization to obtain first standard output, generating old class training samples, inter-class output expressions of the new class training samples between new classes and old classes and intra-class output expressions of the old class training samples between the old classes through a student model, and respectively performing Z score standardization to obtain second standard output and third standard output; calculating inter-class distillation loss according to the first standard output and the second standard output, calculating intra-class distillation loss according to the first standard output and the third standard output, and obtaining total loss according to the inter-class distillation loss, the intra-class distillation loss and the classification loss; and optimizing the total loss of the image classification model to obtain a trained image classification model, and carrying out image classification by adopting the trained image classification model. By adopting the method, the accuracy of image classification can be improved.

The invention belongs to the technical field of federated learning, and particularly relates to a decentralized federated learning method and system oriented to a computing power network. The method comprises the steps that a model demand side in a computing power network uploads a federated learning task to a block chain network, screening computing power nodes according to a reputation score and a data score, and taking the screened computing power nodes as computing nodes to participate in training; the trained local model is verified through the block chain network and then is sent to a model demand side; and the model demander uses a weighted aggregation algorithm based on data fairness to aggregate the local models received in the current round and updates the global model by using an aggregation result. According to the method, the block chain is combined with federated learning under the background of the computing power network, and the universality of the model and the fairness of data are improved while the availability of the federated learning model and the node training enthusiasm are ensured.

The invention provides a post-earthquake emergency material distribution optimization method considering distribution fairness and demand urgency, and relates to the technical field of emergency logistics multi-target distribution optimization under the background of urban earthquake disasters, and the method comprises the following specific steps: S1, determining a basic hypothesis: considering the actual scene of disaster rescue operation, and making the basic hypothesis for the emergency material distribution problem; s2, establishing an emergency material distribution fairness model; s3, establishing a disaster point emergency material demand emergency degree evaluation index system; s4, establishing a multi-target path optimization model; and S5, solving the established multi-target model by using an improved mixed multivariate universe optimization algorithm. The mixed multivariate universe algorithm based on differential evolution is provided, higher convergence precision and better robustness and uniformity are achieved, the earthquake disaster emergency logistics distribution path optimization accuracy is improved, material demand urgency meeting and distribution fairness are guaranteed, the distribution cost is reduced, and the distribution efficiency is improved. And the disaster victim satisfaction degree and the demand emergency adaptation degree are improved.

The present invention discloses a method, device, medium and equipment for multi-department duty scheduling. The method first sorts the duty personnel by department to obtain a duty personnel sequence, and then arranges them into segmented time according to the cyclic order of the duty personnel sequence to form a duty scheduling table. When sorting by department, first sort according to the number of duty personnel in the department, and then calculate the insertion sequence number. The ranking position of the duty personnel in the duty personnel sequence is obtained by calculating the insertion sequence number, and finally the duty personnel sequence is formed. The method disperses the positions of the duty personnel of each department in the duty personnel sequence, so that the duty personnel of each department are averaged in the duty personnel sequence, thereby making the duty scheduling more scientific and fair.

The present invention belongs to the technical field of autonomous driving vehicles, and specifically, is a method for real-time evaluation of the complexity of continuous test scenarios of autonomous driving vehicles. The method comprises the following steps: step 1, establishing a scenario complexity evaluation framework; step 2, analyzing the feature weights of scenario elements; step 3, calculating the element-system complexity mapping relationship; step 4, synthesizing the scene complexity features obtained in steps 2 and 3 to obtain the real-time complexity evaluation results of the continuous test scenario. The present invention can evaluate the complexity of the surrounding environment in real time according to the vehicle driving process, and solves the non-real-time problem existing in the existing test scenario complexity evaluation method.

The invention discloses a Fluent-based numerical simulation and experimental design method for shaking of liquid in a tank of a tank truck, and belongs to the technical field of numerical simulation of fluid dynamics (CFD). The method includes the steps that the hole opening position of the wave-proof plate serves as an independent variable, multiple hole opening position schemes are designed, liquid shaking related indexes are determined to serve as dependent variables, a geometric model is established according to the actual structure of a tank truck tank body, a fluid mechanics model and boundary conditions are set, a solver is configured, output parameters and numerical simulation initial values are set, and an experiment is executed. The liquid shaking process of a certain swash plate trepanning scheme is simulated, acting force data, namely shaking index data, between liquid shaking and the inner wall of the tank body is recorded, the advantages and disadvantages of each trepanning position scheme are evaluated based on the shaking index data, and the optimal scheme is determined. According to the method, a real object experiment is replaced by CFD numerical simulation, parameter setting is carried out, the experiment cost and the experiment risk are remarkably reduced, the experiment efficiency is improved, and data support can be provided for industry design.

The invention relates to a block chain and proxy re-encryption-based outbound data sampling inspection supervision method. The method comprises the steps of generating a first public and private key pair of a data outbound enterprise and a second public and private key pair of a supervisor according to public parameters; encrypting the plaintext data by using the first public and private key pair; carrying out sampling inspection on the first ciphertext obtained by encryption to obtain a second ciphertext, and sending sampling inspection information obtained according to the second ciphertext to the block chain node; after the re-encryption key passes verification, the block chain node re-encrypts the second ciphertext by using the generated re-encryption key; decrypting the obtained third ciphertext to obtain plaintext data; and checking the plaintext data by using the declaration record, and uploading an obtained spot check result to the block chain node. According to the method, the risk of large-scale privacy disclosure is avoided, the supervision efficiency and the supervision effectiveness and reliability are improved, spot check information can be provided as evidence, the transparency of a supervision process is improved, and the trust between an enterprise and a supervision department is enhanced.

This invention relates to the field of satellite communications and addresses the problem of low resource utilization caused by existing hopping beam technology, which only considers the resource allocation of a single satellite and intra-satellite co-channel interference. The method analyzes intra-satellite and inter-satellite interference between two LEO satellites, sets cell weight priorities, and adaptively allocates power based on these priorities to obtain a cell power allocation matrix. Using the average traffic satisfaction of the cell as the optimization objective, the method first considers intra-satellite interference by designing the hopping beam pattern for a single LEO satellite, obtaining the time allocation matrix for each satellite throughout the hopping beam period. Secondly, considering inter-satellite interference, the method rearranges the obtained time allocation matrix using a heuristic algorithm to obtain the final required time slot scheduling matrix. This achieves the avoidance of intra-satellite and inter-satellite interference, significantly improving system throughput while ensuring fairness in user service.

The present application relates to a resource request control method and device for an Internet of Vehicles. The method is applied to a target vehicle, and the target vehicle is any vehicle in the Internet of Vehicles system. The method includes: obtaining the fairness index currently corresponding to all vehicles in the Internet of Vehicles system, and obtaining a set of state parameters of all vehicles in the Internet of Vehicles system; inputting the fairness index and the set of state parameters into a rate control model obtained from a server corresponding to the Internet of Vehicles system to obtain a rate adjustment strategy output by the rate control model; and adjusting the current resource request rate for sending resource requests to the Internet of Vehicles system according to the rate adjustment strategy. In this way, channel congestion can be avoided while ensuring the fairness of minimizing channel congestion, thereby enabling timely communication between vehicles, improving driving safety, and enhancing user experience.

The application discloses a kind of based on big model's exclusive team performance portrait generation and intelligent evaluation system, belong to artificial intelligence and performance management technical field.The system includes data acquisition module, index system management module, performance calculation engine module, portrait generation module, big model intelligent analysis module and visual display module.Index system adopts four-level structure, including six first-level indexes, twenty-four second-level indexes, fifty-eight third-level indexes and a number of fourth-level indexes.Performance calculation engine is based on difference ratio method to calculate index score, portrait generation module generates radar chart, star rating and ranking, and big model intelligent analysis module is based on big language model to carry out intelligent evaluation and ability analysis.The application realizes the scientificity, intelligentization and visualization of exclusive team performance management, and provides strong support for exclusive team construction decision.

The embodiment of the application discloses a method for realizing automatic measurement of long jump, which comprises the following steps: first, using a video acquisition device to acquire a long jump video; determining a long jump area in the long jump video, wherein the long jump area contains a plurality of calibrated measuring points; determining a to-be-detected person in the long jump video; determining a current state of the to-be-detected person according to a current area and a motion posture of the to-be-detected person; when the to-be-detected person is in a measuring state, acquiring a performance measuring point of the to-be-detected person; and finally, determining a long jump performance of the to-be-detected person according to the performance measuring point and the plurality of calibrated measuring points. The method provides a light, convenient, more accurate and stable long jump measurement technology, which does not need to be measured by manual operation, directly acquires a video to realize self-recognition and determination, thereby obtaining an accurate long jump performance, and can also be played back and viewed for multiple times, thereby effectively ensuring fairness and accuracy of the long jump performance.

The application discloses a lossless and efficient data processing method and network interface card based on RMDA, and belongs to the field of data communication of a data center. When a large-scale QP is concurrent in the RMDA, the efficient scheduling processing problem of different QP requests is solved, the lossless request signal, the out-of-order QP request and the waste-free scheduling period are ensured, the scheduling fairness problem of different QPs is effectively solved in combination with congestion control, the blocking of other QP processing by a large message and the excessive number of requests needing to be processed in a single scheduling are avoided, and the head blocking problem is solved.

The application relates to an OFDMA system resource allocation optimization method based on limited channel state information and belongs to the technical field of wireless communication. The research object of the patent is a single base station multi-user OFDMA downlink transmission system, aims to solve a mixed integer nonlinear programming problem with high coupling of optimization parameters, improve the system throughput under the premise of meeting the user fairness, and propose a joint resource allocation optimization algorithm. Under the premise of ensuring certain user fairness, higher system throughput can be achieved, and the application has a wider application range in actual scenes.

The application discloses a kind of configuration methods of active resources, device, storage medium and computer equipment. Among them, the method comprises: server is set to at least two target position information selected first stage activity, initiates first stage activity to multiple clients, receives the first mark information sent by multiple clients through first stage activity, when reaching the mark deadline time corresponding to first stage activity, the mark result corresponding to first mark information is counted, the mark result corresponding to first mark information is determined according to the resource distribution information that activity resource is respectively configured to at least two target position information, according to resource distribution information, the order information that order is matched from multiple orders as target order with at least two target position information is screened respectively, and the configuration task of activity resource is created for target order. The above-mentioned method effectively improves the utilization rate of activity resource, and reaches the expected effect of the configuration of activity resource.

The application discloses a kind of cloud examination room control method, device, equipment and storage medium, the method includes: determining and the same target test question of pre-set standard test question difficulty and the content different target test question, and target test paper is generated based on the target test question;Monitor the behavior state when examinee answers the target test paper;According to the behavior state, it is judged whether the examinee exists cheating abnormal behavior, and according to the judgment result and the answer condition of the examinee answering the target test paper, control is carried out to cloud examination room.Through the above-mentioned mode, target test question is determined and target test paper is generated based on target test question, the behavior state when examinee answers is detected and it is judged whether there is cheating abnormal behavior, and then according to the judgment result and the answer condition control, compared with the existing examination room control method, the above-mentioned mode can accurately judge the behavior of suspected cheating in examination room, and can generate the test paper of different content and the same difficulty, further ensure the fairness of examination.

The application relates to the technical field of shared charging piles and discloses a data credible transaction method for a shared charging pile system. The method comprises the following steps: a shared charging pile data credible model is constructed; when a charging order is monitored to occur, relevant charging data of a charging control terminal, a blockchain node and a mobile terminal of an electric vehicle owner are loaded into the shared charging pile data credible model; the charging data is divided into N data blocks and a secret key is generated step by step, and the hash value corresponding to each data block is calculated; based on the shared charging data credible model, the charging control terminal and the mobile terminal of the electric vehicle owner generate a charging data credible transaction smart contract, the blockchain node exchanges and verifies the secret key, and bill information is synchronized to other blockchain nodes. The application can guarantee the safety and credibility of charging transaction data of the shared charging pile, the privacy data such as the charging transaction is digitally signed to realize anonymous protection, and the data privacy and integrity during the shared charging process and transaction are guaranteed.

The invention discloses a multi-tenant container cloud-oriented RDMA performance isolation system. A monitoring module comprises an eBPF monitoring program and a global context manager; the control path scheduler is deployed in a user mode library of a tenant and is used for receiving feedback information of the monitoring module and limiting or rejecting the original intonation rate of the control path exceeding a threshold value; and the data path scheduler is used for shaping and scheduling a data path task request issued by the tenant application program so as to control the sequence and the speed of entering the RDMA network card. According to the method, the eBPF monitoring program is adopted to monitor the original intonation frequency of the control paths of all tenants and the resource use condition of the RDMA network card, and millisecond-level response to RDMA control path operation is achieved. Meanwhile, a dynamic task segmentation mechanism based on active delay is adopted, and RDMA network card resource occupation of different tenants is balanced; and fairness is ensured based on a scheduling strategy of admission control.

The invention provides a power-carbon transaction method based on a multi-leader and multi-follower Stackelberg game, and the method comprises the following steps: S1, enabling carbon emission to be embedded into the income of each producer in a currency manner based on the MLMF Stackelberg game, and forming a power-carbon integrated market; s2, for an MLMF game model with the electric power and carbon transaction prices as decision variables, proving the existence and uniqueness of Stackelberg equalibrium; s3, deducing an electric power and carbon transaction price formula under the SE by adopting a centralized analysis algorithm; and S4, expanding the two-stage transaction to the electric power-carbon integrated market. According to the invention, low-carbon operation is realized, fairness of a transaction market and security of a network are ensured, carbon emission is significantly reduced, and the total income of producers and sellers is not damaged.

The utility model relates to the technical field of mahjong competitions, and discloses a weak-touch mahjong tile and a mahjong combination thereof, the weak-touch mahjong tile comprises a mahjong tile body, two corresponding side surfaces of the mahjong tile body are respectively a mahjong tile back surface and a mahjong tile front surface with patterns, and a height difference exists between the patterns and the mahjong tile front surface. The range of the height difference is 0.1 to 0.6 mm; the mahjong combination comprises a mahjong box and the mahjong tiles, the mahjong box comprises a box body and a box cover, and a tile touching opening for taking the mahjong tiles is formed in the box cover. In practical application, the mahjong tiles are simpler and more convenient in manufacturing process and lower in cost while mahjong games do not need to be stacked and shuffling time is shortened.

This invention relates to the field of concurrent data processing technology in embedded operating systems, and particularly to a hierarchical fair spinlock control method based on queue distance awareness. The method includes the following steps: the processor initiates a mutual exclusion access request for shared resources; reads the idle / available flag stored in the lock; if the flag is valid, the processor directly acquires the lock through an atomic comparison and exchange operation; if the flag is invalid, the processor atomically increments the pending ticket number stored in the lock to generate a ticket number held by the processor; reads the current service ticket number stored in the lock, and calculates the difference between the processor-held ticket number and the current service ticket number to obtain the queue offset. This invention, while maintaining the fairness and determinism of spinlocks, enables safe execution in various modes including tasks, interrupts, and exceptions, while reducing cache consistency overhead in multi-core concurrent scenarios, improving lock scalability, and providing lock state detection capabilities.

The invention discloses a big data platform resource processing method and equipment in a multi-tenant environment, mainly relates to the technical field of resource processing, and is used for solving the problem that a conventional resource allocation method mostly adopts static quota or a simple oversale strategy; and the method is difficult to adapt to complex scenes with dynamic and changeable business of the tenants and obvious resource demand difference. Comprising the steps that platform resources distribute a basic quota for each tenant, and the remaining platform resources serve as a shared resource pool; calculating a historical fairness score according to the matching degree between the actual resource utilization rate of the tenant in the preset time period and the basic quota; calculating a total score of the tenants based on the business priority, the historical fairness score and the resource contribution degree; according to the descending order of the total scores of the tenants, determining the order of obtaining the elastic quotas from the shared resource pool; and determining a specific resource scheduling strategy of the tenant about the basic quota and the elastic quota according to the service scene.

The present application relates to the technical field of park energy data management, and particularly relates to a zero-carbon smart park comprehensive energy method and system based on big data, wherein the method comprises the following steps: S1, generating a carbon equity unit record; S2, establishing a corresponding source summary for each carbon equity unit; S3, writing the identification information corresponding to the transferred carbon equity unit into the restriction identification set corresponding to the first node; S4, generating a merged source summary; inheriting the source summary before splitting; S5, performing backflow risk verification based on the source summary attached to the to-be-transferred carbon equity unit and the identification information in the restriction identification set; S6, using the registration result for comprehensive energy scheduling and carbon accounting; the method effectively avoids the loss of source information and the risk of repeated offsetting caused by merging and splitting, and facilitates linkage with comprehensive energy scheduling and carbon accounting, thereby improving the real-time performance, implementability and system autonomy of zero-carbon smart park carbon equity management.

The application relates to a reinforcement learning congestion control method based on time series modeling of a Transformer, and belongs to the technical fields of communication networks and artificial intelligence, and comprises the following specific steps: step 1: network simulation is performed at an NS3 end to generate simulated network data; step 2: the generated simulated network data is sent to a simulated TCP-RL protocol, the TCP-RL protocol simulates the sending of data packets, delay, packet loss and the return of ACK; step 3: a congestion control RL environment at a python end; the simulated network data sent by the NS3 end is received, is converted into observation data and is sent to a reinforcement learning intelligent agent; step 4: the reinforcement learning intelligent agent at the python end analyzes the received observation data, forms a history sequence, dynamically pays attention to the real-time situation of data sending and returns the congestion control result of analysis and decision to the simulation network environment of the NS3 end as an action in reinforcement learning. The application can make the congestion control effect more optimal and more stable compared with a traditional algorithm.

The application discloses a fair personalized recommendation method and device based on mutual information decoupling and a storage medium, and steps of the method comprise the following steps: 1. constructing original data, including a user-product rating matrix and a user sensitive attribute matrix; 2. constructing a sensitive embedding network to learn sensitive embedding of a user and a product, including a biased one-hot encoding layer, a sensitive information encoder and a sensitive attribute prediction layer; 3. constructing a hybrid embedding network to learn hybrid embedding of the user and the product, including a hybrid one-hot encoding layer, a hybrid information encoder and a preference prediction layer; and 4. constructing a non-sensitive embedding network to learn non-sensitive embedding of the user and the product, including an unbiased one-hot encoding layer, a mutual information lower bound optimization layer and a mutual information upper bound optimization layer. Through the double mutual information fairness constraints on the embedding vectors, the application improves the fairness of the recommendation system while ensuring the recommendation accuracy.

The invention discloses a hybrid multiple access method suitable for a low-orbit satellite Internet of Things, and the method comprises the steps: firstly, enabling a low-orbit satellite and an unmanned plane to obtain the position information of ground Internet of Things equipment; secondly, on the basis of the obtained position information, calculating a position correlation coefficient between the devices, and performing grouping optimization on the multiple pieces of non-direct-transmission Internet of Things devices; then, the unmanned aerial vehicle relay serves one non-direct transmission device group in each time slot by adopting a method of combining polling scheduling with a beam forming technology based on position information; and finally, a plurality of direct transmission Internet of Things devices share spectrum resources with the unmanned aerial vehicle, a multi-beam non-orthogonal multiple access technology based on weighted beam superposition is adopted to access a satellite network at the same time, and the hybrid multiple access method provided by the invention can make full use of resources such as a time domain, a space domain and a power domain. The heterogeneous flexible access requirements of different types of Internet of Things equipment are effectively met, and the method has the advantage of low algorithm complexity, so that the system overhead is remarkably reduced, and the method is suitable for a low-orbit satellite Internet of Things access scene.

This invention relates to a detection method and system for controlling charging piles, relating to the field of charging piles. The method includes: in response to activation information upon installation completion, acquiring detection distance information and determining the installation scenario based on the detection distance information, the installation scenario including indoor and outdoor scenarios; acquiring environmental information and user physical characteristics within a preset detection range based on the installation scenario; analyzing the user physical characteristics based on weighted feature fusion judgment rules to determine user characteristics within a preset usage range; judging the user characteristics using a preset judgment method to determine the user type; matching a control strategy based on the user type; establishing an environmental model using the environmental information within the detection range, and generating dynamic adjustment parameters based on the environmental model; and adjusting and controlling the charging pile according to the control strategy and using the real-time dynamic adjustment parameters. This application has the effect of improving the safety of charging pile use.

The utility model provides a kind of card face reconstruction device of competitive egg smashing contest, comprising: table mat, screen, playing cards, two displays, data processor and movable window board, screen is vertically arranged on the table mat, screen has a hollow accommodating space, a window is equipped in the center below screen, the size of movable window board is slightly greater than window, and movable window board is movably arranged in accommodating space and at least has two states of shielding window and opening window, table mat is equipped with the card area, located in the two sides of window, the card area is equipped with identification module, playing cards are equipped with readable label, the readable label of playing cards is equipped with identification ID, identification ID is bound with the card face information of card.The card area of the utility model can automatically identify card face information and transmit it to display through data processor, avoid the risk of illegal information transmission between teammates, increase the fairness of competition.

The application relates to a parking attendant performance evaluation method for a smart parking management system and belongs to the technical field of parking management. In the method, the historical difference value between the historical payment time and the corresponding historical departure time in all historical parking payment orders is calculated, a time difference data set composed of all the historical difference values is input into a statistical analysis model for time distribution analysis, a time evaluation window with different correlation strengths with the parking attendant is generated according to the output distribution inflection point and a preset business logic, a corresponding weight coefficient is allocated to each time evaluation window through a dynamic weight calculator, the current difference value between the order payment time and the vehicle departure time in all parking performance attribution records of the parking attendant is matched with the time evaluation window, so that the weight coefficient corresponding to each parking performance attribution record is obtained, and the performance evaluation result of the parking attendant is generated in combination with multidimensional indexes. The application guarantees the fairness of the performance evaluation of the parking attendant and improves the working enthusiasm of the parking attendant.

The invention discloses an intelligent control blue code software system for online and offline sales transaction of a store entity enterprise, and belongs to the technical field of electronic commerce. The invention provides an intelligent control blue code software system for online and offline sales transaction of a store entity enterprise, which can improve the operation efficiency of a merchant, reduce the operation cost, improve the shopping experience of consumers, encourage consumption through a point reward mechanism, realize automatic processing of transaction, improve the transaction efficiency, and improve the user experience. The fairness and normalization of the system are ensured, each business license only limits one merchant to enter, and the system can help the merchant to improve the sales efficiency, reduce the operation cost and improve the shopping experience of consumers through intelligent management and control.