10 results about "Association strategy" patented technology

The application provides an explainable sequence recommendation method, system and device fusing concepts and behaviors, relates to the technical field of recommendation systems, and the method comprises the following steps: obtaining a concept library and a historical behavior sequence of a user; obtaining a historical concept sequence and a concept embedding matrix according to the historical behavior sequence and the concept library; obtaining a concept weight matrix and a concept behavior sequence set according to the historical behavior sequence, the historical concept sequence and the concept embedding matrix in combination with an association strategy of concepts and behaviors; and performing sequence recommendation on the user according to the historical behavior sequence, the concept embedding matrix, the concept weight matrix and the concept behavior sequence set to obtain an explainable sequence recommendation result. By introducing the association strategy of concepts and behaviors, the application not only effectively fuses abstract concept information and specific behavior information, but also dynamically adjusts the understanding and explanation of user preferences through the concept embedding matrix and the concept weight matrix, thereby improving the accuracy and explainability of the sequence recommendation result.

This application relates to a method for designing a long-term stable association strategy for a digital twin edge network. The method includes: constructing a long-term association strategy model and a communication and computing model for the digital twin edge network; constructing decision variables and introducing these variables into the long-term association strategy model and the communication and computing model to construct a problem minimizing system latency; decoupling and optimizing the problem minimizing system latency to obtain the decoupled optimized problem; and solving the decoupled optimized problem to achieve long-term stable association of the digital twin edge network. This invention considers the long-term cumulative benefits of the network system, systematically considers state evolution and resource allocation across time periods, and optimizes overall performance from a global perspective. Compared to short-term association strategies, long-term stable association strategies can better balance immediate benefits and future costs, reduce the long-term average latency and average energy consumption of the system, and are particularly suitable for network system optimization in dynamic and uncertain environments.

According to the database lease-driven exclusive task anti-re-execution method provided by the invention, by introducing a dynamic evaluation and updating mechanism of the effectiveness of the association strategy, the intelligence and environmental adaptability of exclusive task anti-re-execution are effectively improved, the effectiveness of the historical strategy in the current system state can be accurately evaluated, and the reliability of the system is improved. And an optimal anti-re-verification strategy is generated in real time based on multi-objective optimization, so that the risk of task repeated execution or error interception is remarkably reduced. Compared with a static configuration scheme, the method has the advantages that collaborative adaptive adjustment of key parameters of the system is realized, and the system can actively respond to dynamic environments such as load fluctuation and network change. The dynamic relevance ensures the data consistency and the system reliability in the processing of high-priority tasks such as electronic certificates, optimizes the resource utilization rate, reduces the performance bottleneck or resource waste caused by strategy lag, and provides more efficient and reliable anti-repetition guarantee for distributed task scheduling.

The invention discloses a 3D target trajectory automatic labeling method for point cloud denoising and trajectory verification, which comprises the following steps: firstly, performing data enhancement on continuous point cloud data, and combining past point cloud frames with current point cloud frames by using time information to obtain various point cloud frame combined data; then, a 3D target detection algorithm is adopted to obtain 3D target detection results of the point cloud frame combination data, weighted fusion is carried out, a point density perception voxel network is utilized to carry out adjustment and filtering, trajectory tracking is carried out based on data observation and an association strategy, a forward initial 3D target trajectory set and a reverse initial 3D target trajectory set are generated, and the forward initial 3D target trajectory set and the reverse initial 3D target trajectory set are combined; a forward 3D target track set and a reverse 3D target track set are obtained through track legality verification, and weighted fusion processing is carried out to obtain a 3D target track set; the method has the advantages that the labeling efficiency is relatively high, the labeling cost is relatively low, and meanwhile, the automatic labeling of the 3D target trajectory can be completed completely and accurately.

The invention relates to the technical field of advertisement putting, in particular to a crowd portrait method and promotion system based on big data, and the method comprises the following steps: S1, data collection and screening: screening the collected data according to an input instruction, and classifying the screened data according to requirements; s2, establishing an association strategy, and marking persons belonging to the same group of data according to the association strategy; and S3, iteratively updating the marks according to the use data. Through the step S3, definition tags of individuals (persons) to which the data belong are continuously updated and refined, so that the specified data flow to the specified crowd is greatly facilitated, and the accuracy of the data flow is greatly improved. And a large amount of dynamic data is gathered into label words one by one in a marking manner, so that the amount of data analysis is greatly reduced, and the method is suitable for large-scale crowds.

The invention relates to the technical field of multi-target tracking, in particular to a motion perception multi-target tracking method based on query transfer, which comprises the following steps: receiving visual target query and multi-source data stream; a combined expert decoder is adopted to process the visual query, and visual behavior characterization capable of adapting to complex motion of the target is generated; performing time sequence coding on the multi-source data stream to generate a data sequence; a time sequence consistency reward value is generated by calculating the overlapping degree of the visual residence time sequence and the data active time sequence, and reinforcement learning training is carried out on the association strategy model according to the time sequence consistency reward value; inputting the fused state space into the model to generate an association decision, and constructing a global identity trajectory; and when the target is reproduced, rebinding is executed through a multi-dimensional matching mechanism fusing appearance, motion and space prior. Through fusion of a visual trajectory and multi-source Internet of Things data and introduction of a reinforcement learning optimization long-term association strategy, accurate and long-term cross-modal target tracking in a complex farm interaction scene is realized.

The invention relates to the technical field of artificial intelligence, and discloses an artificial intelligence-based text emotion analysis and emotion recognition system, which comprises a text acquisition module, an emotion feature module, an association strategy module, a context optimization module, a dynamic reference module and an emotion decision module. The text collection module obtains a multi-source heterogeneous text and sets a semantic analysis interval; the emotion feature module divides emotion nodes and generates emotion feature vectors; the association strategy module extracts key indexes to establish emotion recognition rules; the context optimization module dynamically corrects an index through a context association mode and calculates a semantic offset; the dynamic reference module deduces an optimal emotion threshold value to generate an emotion deviation sequence; and the emotion decision module integrates the deviation sequence to form a recognition scheme. According to the system, multi-level analysis of text sentiment is realized through multi-module cooperation, cross-domain adaptability and sentiment recognition accuracy under a dynamic context are improved, and the system is suitable for scenes such as public opinion monitoring and user feedback analysis.

The invention relates to the technical field of data processing, and discloses a data auditing method, computer equipment and a computer program product. The method comprises the steps of obtaining to-be-audited data of multiple application architecture hierarchies, binding the to-be-audited data of the multiple application architecture hierarchies based on a preset association strategy, determining the risk degree of the bound to-be-audited data based on association features of the bound to-be-audited data, and executing a risk avoidance strategy corresponding to the risk degree according to the risk degree. Wherein the plurality of application architecture hierarchies comprise a user layer, a request layer and a database layer, and the preset association strategy represents a mapping relationship of the to-be-audited data among the plurality of application architecture hierarchies. The efficiency and accuracy of the audit evaluation process are improved.

The invention discloses a method for managing resources in a cellular-free network based on deep reinforcement learning, which comprises the following steps of: deploying a deep Q network DQN in a centralized manner on a central processing unit, and optimizing a user association strategy on a large time scale to ensure that a user meets the requirement of minimum spectrum efficiency; a deep cycle Q network DRQN is deployed on each access point in a distributed manner, and beam forming and power distribution strategies are optimized on a small time scale, so that maximum system energy efficiency is realized; and carrying out iterative learning on the deployed two-layer network on two time scales of a time frame and a time slot until convergence so as to find an optimal resource management strategy. According to the method, through a centralized-distributed deployment mode and a large-scale and small-scale updating mode, the information interaction complexity and the calculation complexity are reduced, and joint optimization of user association, beam forming and power distribution is achieved.

The present application belongs to the field of computer vision and multi-target tracking, and relates to a multi-target tracking algorithm for a target nonlinear motion scene. The present application improves the detection feature and target re-identification feature extraction mode of the existing multi-target tracker, and at the same time, in order to be more suitable for the characteristics of target-track association matching under the nonlinear motion condition, a new association strategy is proposed in the specific matching association stage. The present application improves the calculation cost, and improves the problem that the existing multi-target tracker is difficult to associate targets under the pedestrian nonlinear motion scene, and improves the tracking effect of the multi-target tracker under the pedestrian nonlinear motion scene.