14 results about "Information gap" patented technology

This application discloses a data processing method relating to the field of artificial intelligence, comprising: obtaining a first query corresponding to a first neural network and a second query corresponding to a second neural network; the first query is a query obtained by maximizing the information entropy of the query corresponding to the first neural network; the first neural network is a model obtained by compressing the second neural network; the first neural network and the second neural network are used for object detection; determining a first loss based on the first query and the second query; the first loss indicates minimizing the information difference between the second query and the first query; and updating the first neural network and the first query based on the first loss. This application improves the model accuracy of the compressed model by minimizing the information gap between the query queries of the student model and the teacher model, thereby minimizing the information gap between the student model and the teacher model.

This application discloses a code generation method, apparatus, device, medium, and product, relating to the field of application development. The method includes converting front-end application design data into a structured dataset containing page element information, initial interaction topology, and visual design specifications, addressing the problem of insufficient interaction logic description capabilities. It then converts this data into an attributed state transition diagram, describing the complete interaction attributes of the front-end application's page-to-page interaction from pre-interaction preparation to the completion of the interaction's effect, replacing vague natural language descriptions and achieving machine-analyzable interaction logic, bridging the information gap between design and implementation. Finally, it combines a pre-trained language model to directly generate front-end development code, achieving automated and standardized output from design to code, addressing the lack of machine-analyzable formal representation of front-end interaction logic, thereby improving front-end development efficiency and code quality, and reducing communication and maintenance costs.

This invention provides a simulation experiment planning and decision-making intelligent agent and its method. This agent, through deep fusion of a large language model, achieves end-to-end intelligent generation and dynamic closed-loop optimization of experiment planning schemes. The agent possesses natural language interaction capabilities, automatically generating draft experiment task sheets and proactively inquiring about key details; it provides customized recommendations for equipment and algorithms based on real-time knowledge base queries; it automatically converts experimental conclusions using semantic understanding; and it continuously optimizes the knowledge base through an "execution-feedback-evolution" closed loop, upgrading traditional passive experiments into an agile decision-making system driven by proactive cognition. This invention solves the problems of information gaps, inefficient decision-making, and lack of evolutionary capabilities in simulation experiments, providing accurate, efficient, and sustainable intelligent support for simulation experiment planning, significantly improving the reliability and efficiency of key tasks such as top-level experimental planning, task decomposition, scheme generation, and experiment execution.

The present application belongs to the technical field of comprehensive energy, and discloses a networked hydrogen-based micro-grid multi-period planning method based on probability constraint and information gap decision. A two-stage multi-period planning framework is constructed, a hybrid chance constraint-information gap decision theory (CCP-IGDT) co-modeling method and an adaptive prediction-correction Benders decomposition (APC-BD) algorithm are designed, the robustness of the system to unexpected hydrogen load growth is maximized under fixed budget constraints, and the heterogeneous characteristics of short-term random fluctuations and long-term structural uncertainties are effectively handled.

This invention discloses a multi-machine collaborative method and system for multimodal cargo handling and transshipment, belonging to the field of logistics and transportation technology. By acquiring comprehensive information on transportation tasks and real-time status data of operating equipment, this invention clarifies cargo attributes, timeliness requirements, site constraints, and equipment operating parameters, avoiding deviations caused by data fragmentation during operations. This improves the targeting and feasibility of multi-machine collaborative operations, effectively reducing resource waste and operational delays caused by information gaps. Through a scientifically formulated plan, precise equipment scheduling, real-time process monitoring, and dynamic strategy adjustment workflow, combined with equipment selection, route planning, task allocation, and handover verification, it achieves orderly and efficient execution of multi-equipment collaboration. This ensures precise matching between equipment and tasks, synchronous coordination of the operational process, and rapid response to emergencies such as cargo damage and equipment malfunctions, improving handling and transshipment efficiency and cargo integrity, and enhancing operational resilience and flexibility.

The application relates to the technical field of computers and discloses a software development application data processing method based on an AI large model, which comprises the following steps: adopting unified event extraction measures on an original research and development and delivery data set to generate an associated event carrier; adopting evidence graph assembly and conservative merging constraints on the event carrier to generate a delivery evidence chain corresponding to each release; adopting fingerprint consistency checking and software bill of materials cross verification measures on the evidence chain to generate consistency results and a verified bill of materials; adopting joint risk scoring measures on alarms, dependency changes and consistency results to generate a treatment label; adopting leakage limited processing measures on the evidence chain, the bill of materials, traceability proof and the treatment label to generate a large model available data package; and adopting an evidence cited assembly template and evidence coverage rate gate control measure on large model output to generate an output package and output an information gap list when evidence is insufficient.

A layout optimization method, system, electronic device, storage medium and program product, the layout optimization method comprising: obtaining a conflict pattern group that does not conform to a layout parameter rule in a target layout; within a lithography limit process window, shrinking the feature size of a pattern in the conflict pattern group to an optimized feature size to obtain an optimized target layout. Since the shrinking of the feature size will affect the pitch, the method of adjusting the feature size of the pattern in the conflict pattern group to the optimized feature size within the lithography limit process window improves the process capability limitation problem, and easily makes the optimized target layout conform to the process requirement; moreover, the method of adjusting the feature size of the pattern to improve the process capability limitation problem also reduces the demand for re-drawing the layout, which is also conducive to reducing the probability of repetitive work caused by the information gap between the layout design party and the manufacturer, thereby improving the work efficiency.

The application discloses a cross-role cooperative difference alignment method, device, equipment and medium, and the method comprises the following steps: obtaining role state data of each role based on role agents of multiple roles; comparing the role state data to identify a target role having a cooperative difference with other related roles; pushing verified benchmark cooperative information to a target role agent corresponding to the target role according to the cooperative difference; and guiding the target role to align the difference through the target role agent based on the benchmark cooperative information. The method realizes real-time collection of role state data of each role through the role agent, can accurately capture cognitive bias and information gaps of multiple roles in task execution, and performs directional intervention in combination with benchmark cooperative information verified dynamically, so that the target role can adjust behavior and decision-making according to a unified standard in a complex cooperation scene, and dynamic balance and efficiency optimization of a cross-role cooperative state are realized.

This invention relates to the field of UAV swarm cooperative control technology, and discloses a multi-UAV cooperative path planning method based on dynamic communication quality assessment. The method includes offline pre-training to generate a configuration decision model and a configuration-region partitioning joint mapping table; online detection of sudden interference and extraction of interference feature parameters; rapid selection of target configuration based on the pre-trained configuration decision model and simultaneous acquisition of region partitioning schemes; execution of topology reconstruction and adoption of a double-buffering mechanism to maintain the continuity of field function modeling; and guidance of cooperative path optimization based on the updated field function model. This invention reduces the topology reconstruction decision delay by two orders of magnitude and eliminates the information gap in field function modeling during the topology reconstruction transition period.

The application discloses a kind of mine comprehensive energy system information gap decision robust scheduling method and system, the method includes: using information gap decision theory to establish considering associated resource experience proportion disturbance mine comprehensive energy system information gap decision robust scheduling model;Model includes the uncertainty constraint of associated resource and load demand and associated resource experience proportion disturbance constraint;The double-layer structure model of mine comprehensive energy system information gap decision robust scheduling model is converted into single-layer structure model;Linear conversion is carried out to relevant nonlinear constraint in the mine comprehensive energy system information gap decision robust scheduling model converted into single-layer structure, and mixed integer linear programming model is established;Through the mixed integer linear programming model established, calling solver is solved.The uncertainty of coal associated resource is fully quantified, and the optimization scheduling scheme prepared ensures that mine comprehensive energy system can still reliably operate under the worst scenario possibly existing.

This invention provides a method and system for quantifying and evaluating individual employee workload, belonging to the field of employee workload management technology. By proposing the WU decomposition method for the first time, it classifies and breaks down various complex tasks in the R&D and management processes into multi-level, standardized categories, defining the smallest quantifiable work unit. This transforms abstract work content that was originally difficult to measure into calculable WU values. A comprehensive individual evaluation system covering multiple dimensions such as "workload saturation, workload, work status, work ability / efficiency, work attitude, and overtime" is constructed. This breaks through the limitations of traditional evaluations that only focus on work completion, providing organizations with a comprehensive and three-dimensional data profile of individual employee performance. It not only improves the efficiency of data entry and processing, but more importantly, it bridges the information gap between evaluation data and actual work data, achieving efficient collaboration between work execution and work evaluation.

This invention relates to an intelligent management system, method, medium, and equipment for fuel assembly lifecycle data. A data entry module inputs basic data for the fuel assemblies; a data aggregation module aggregates and sorts the basic data; a data query module analyzes, processes, and integrates the data, outputting different categorized data; and a data statistics module performs aggregated statistics and outputs the results. These modules work collaboratively to achieve hierarchical management of fuel assembly data, lifecycle data management, one-click generation of a list of available new assemblies and detailed information, a list of spent fuel assemblies meeting transportation conditions and detailed information, and detailed information for all fuel assemblies. This invention breaks down data silos and information gaps in traditional management models, establishing accurate, dynamic, and traceable lifecycle electronic files for each fuel assembly. It provides reference for nuclear fuel security, nuclear material balance management, refueling design, spent fuel transportation, and other operations, ensuring the integrity, accuracy, timeliness, and standardization of data.

The invention discloses a communication field retrieval enhancement method based on embedded model fine tuning and self-adaptive iteration, which fully considers semantic particularity of communication terminologies and deep logic association behind complex technical problems, and accurately matches and queries highly related document fragments to enhance the retrieval of the communication field. According to the method, a high-quality background context is constructed and input to a large language model, the retrieval accuracy and question and answer quality of a system are remarkably improved, and under the premise that expensive full-amount LLM fine tuning is not carried out, the understanding and answer accuracy of the large language model in the telecommunications professional field is remarkably improved through low-cost embedded model fine tuning and an intelligent retrieval strategy; according to the invention, the method can process highly specialized technical documents and complex queries, has a deep semantic understanding capability for domain-specific terms, and can actively perceive and fill up information gaps through a multi-round iterative retrieval mechanism, so that an intelligent question-answering system is promoted to develop in a more professional and more accurate direction.

The application provides an AI active care intelligent interaction method based on long-term memory and multi-round dialogue, and belongs to the technical field of natural language processing.The application realizes fine-grained distinction of the importance of information in long dialogue history by constructing a weighted memory tensor based on an exponential decay mechanism, retains complete interaction tracks and highlights recent semantic contributions, and effectively solves the long dependence information loss problem caused by fixed window truncation; through deep fusion of entity embedding, weighted features of the appearance round and dialogue theme evolution path, joint encoding with timing perception ability is generated and stored in a session layer dictionary, which significantly enhances the system's ability to capture user interest migration; further, through a dynamic window detection mechanism, semantic missing areas in the dialogue are automatically identified and memory enhancement based on linear interpolation is performed, filling the information gap and improving the coherence and pertinence of active questioning.