80 results about "Error reduction" patented technology

The invention relates to the technical field of intelligent question answering methods, in particular to a case question answering method based on a large language model, a medium and equipment. The method specifically comprises the steps of obtaining graphic and text information and related questions of a case; and inputting the intelligent question and answer model to obtain an optimal answer. The intelligent question and answer model is obtained through the following steps: training a pre-training model through a constructed multi-modal corpus based on transfer learning to obtain a base model; based on LoRA and P-Tuning, performing fine tuning on the parameters of the base model to obtain a fine-tuned large language model; introducing a CLIP module to perform feature fusion on the graphic and text information to obtain a preliminary question and answer model; introducing a thinking chain to obtain a generation model capable of outputting a plurality of candidate answers with reasoning chains; and training an existing scoring model to obtain an evaluation model capable of outputting an optimal candidate answer. According to the method, the performance of the question answering technology in the aspect of generating accurate and reliable answers can be improved, and error information or illusion is reduced.

The invention provides a robot control method and system based on a big language model, a double big language model evaluation mechanism is adopted, a big language model of a code generation and modification module uses Claide 3.5 sonnet, control codes are generated by the Claide 3.5 sonnet, evaluation is provided by adopting a gpt-4o evaluation model of OpenAI to optimize suggestions, and the big language model of the code generation and modification module is used for evaluating the control codes. The risk that problems are difficult to find due to homogenization of the same module is reduced, and the generation quality, safety and supportability are improved; in addition to a double large language model evaluation mechanism, due to the fact that safety prompt words are added into prompt words of a generation model and a generation template example is provided, operation of the generation model in a specific framework can be limited, the possibility of error occurrence is reduced, and specification and safety of a generation effect are further provided; in addition, the control class verification module constructs a simple simulation environment after the control class codes are generated, so that grammar errors and logic vulnerabilities of the generated codes can be screened in advance, and potential safety hazards are further reduced.

The invention discloses a multi-frequency multi-mode signal multi-path error weakening method, and relates to the technical field of precision positioning, and the method comprises the steps: constructing a multi-frequency multi-mode non-combination precision single-point positioning observation equation, separating multi-path errors in pseudo-range and phase observation values through the equation, and forming residual data; performing quality control on the residual error, and eliminating an abnormal value; feature extraction and judgment are carried out based on residual data, wherein time repeatability features, spatial distribution features and trend complexity features are extracted; selecting a correction model according to a feature judgment result, wherein the correction model comprises a fixed star daily filtering model, a multi-path hemispherical graph model and a trend surface analysis multi-path hemispherical graph model; and generating a multi-path error correction amount through a selected model, correcting an original observation value, and outputting a positioning result. The method solves the problem that the positioning precision is reduced due to multi-path errors, can effectively weaken the multi-path errors, improves the positioning precision and stability, adapts to a complex environment, and reduces data processing redundancy.

The invention provides an intelligent contract examination and generation system based on an AI large model. The intelligent contract examination and generation system comprises the following steps: S1, data processing; s2, constructing a knowledge base; and S3, model training architecture. According to the intelligent contract examination and generation system based on the AI large model, the seriousness requirement of laws and regulations is met, the accuracy of the AI model in the law field can be greatly improved through close cooperation with a law expert team and fine adjustment of mass data, errors and illusion are reduced, the law compliance of the contract is concerned, and the law compliance of the contract is improved. More attention is paid to specific requirements and sites of users, customer interests are protected, contract terms better conforming to the user interests are generated by a tool by analyzing service backgrounds and requirements of the users, user interface design is simple and visual, the users can easily use the contract terms without professional legal knowledge, the users only need to describe the requirements through natural languages, and user experience is improved. Therefore, the contract can be examined and generated, the use threshold is greatly reduced, and the user experience is improved.

The embodiment of the invention discloses a switch configuration management method and device, equipment and a medium. The method comprises the following steps: acquiring a switch configuration requirement input by a target object; wherein the switch configuration requirement is described based on a natural language; performing keyword analysis on the switch configuration demand by using a large language model to obtain an analysis result, and determining a plurality of target query statements according to a preset generation mode and the analysis result; wherein the preset generation mode is determined based on a database type; querying the plurality of target databases according to the target query statement to obtain a query result, and determining a target configuration command according to the query result; and sending the target configuration command to the target switch, and executing the target configuration command through the target switch to obtain an execution result. According to the technical scheme, the target configuration command is determined based on the multi-database query result by using the large language model, so that the error rate can be effectively reduced, the configuration accuracy and safety are ensured, and the intelligent level of network management is improved.

The invention discloses a power grid frequency balance intelligent control system based on a distributed architecture, relates to the technical field of power grid control, and is used for solving the problems that the existing WLS / UKF and other methods are generally assumed to be good in synchronization, are insensitive to heteroscedasticity time delay and missing measurement, and are difficult to ensure robustness under real communication and equipment conditions. Through a through link, under the real clock misalignment, time delay jitter and packet loss conditions, measurement of a unified time axis and quality quantification are realized, and frequency estimation bias and uncertain transmission are significantly reduced. Reconstruction, filtering and triggering are penetrated with the same confidence degree, and abnormal measurement suppression and effective information utilization are improved; the fusion efficiency and timeliness of asynchronous arrival data are improved by utilizing graph prior and information domain incremental updating; and regional measurement abnormity and real power imbalance are discriminated through double-model causality, and error event reporting is reduced.

Systems and techniques that facilitate quantum noise suppression are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory that can execute the computer executable components stored in memory. The computer executable components can comprise an error detection component that determines one or more portions of a quantum circuit susceptible to noise errors; and an error reduction component that compiles an inverse of the noise errors into the quantum circuit based on context of the quantum circuit and the one or more determined portions of the quantum circuit.

Systems and techniques that facilitate quantum noise suppression are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory that can execute the computer executable components stored in memory. The computer executable components can comprise an error detection component that determines one or more portions of a quantum circuit susceptible to noise errors; and an error reduction component that adds one or more dynamical decoupling sequences to the one or more determined portions of the quantum circuit wherein the one or more dynamical decoupling sequences are determined based on context of circuit layers of the quantum circuit.

The present invention provides a computer system and a computer-implemented method for adaptive quality management in artificial intelligence (AI) processing. The system comprises a memory unit storing machine-readable instructions and a processor configured to execute the instructions to store, in a persistent correction memory, records of prior corrections to AI outputs; retrieve a correction for a current input based on similarity between the input and stored records; generate, by an audit-trail module, a tamper-evident record of the correction and its application; schedule, by a predictive resource manager, AI workloads in dependence on service-level constraints, carbon-intensity data, and quality metrics; and update, by a feedback controller, retrieval and scheduling parameters based on effectiveness information. The processor thereby enables the persistent correction memory, audit-trail module, predictive resource manager, and feedback controller to operate as a closed-loop adaptive system that reduces recurrence of AI errors, ensures verifiable auditability, and optimizes computational efficiency.

The invention discloses an automatic driving vehicle accumulative error reduction method based on Kalman filtering, and aims to improve the precision of vehicle trajectory prediction and reduce accumulative errors. The system obtains a first attitude information vector phi k1 and a second attitude information vector phi k2 of a vehicle through a wheel encoder and a vehicle-mounted GPS. And performing prediction calculation on the two attitude vectors according to the constructed Kalman filter to obtain a corrected optimal attitude information vector phi opt. And substituting the optimal attitude information vector phi opt into the starting and stopping conditions of the automatic driving vehicle, and comparing the distance obtained by calculation with a set threshold value, so as to perform judgment and calibration according to the size of an error. A simulation test shows that the actual position data of the vehicle is calibrated by the steps when the vehicle is parked each time, so that the accumulative error of the vehicle in a multi-time starting and stopping mode is reduced, the system effectively reduces the error superposition of each section of track, and the accuracy of predicting the long-distance multi-section driving track is improved.

The invention relates to an input adaptive precision Cordic algorithm implementation method and device. The method comprises the following steps: obtaining truth values of input x and y and comparing the truth values; performing most significant bit estimation on the truth value with the larger truth value, and obtaining a corresponding shifting mode; shifting the truth values of the input x and y based on the queried shifting mode; and inputting the complemented input x and y into the Cordic operation unit. According to the invention, the spare bit width in the input data is fully utilized to reduce or avoid the occurrence of error iteration, so that the occurrence of error iteration can be reduced to the greatest extent and the accuracy of the output result can be improved.

The invention discloses a small-probability multi-failure mode system reliability analysis method fusing ICE-SYS and MRGP. The small-probability multi-failure mode system reliability analysis method comprises the following steps: S1, generating an initial sample construction training set; s2, constructing a multi-response Gaussian process MRGP model according to the training set, executing a cross entropy important sampling method ICE-SYS of the system on the MRGP model, and taking the last layer of samples of the ICE-SYS as a candidate sample pool S; s3, calculating a system prediction response mu * (x) and a prediction standard deviation sigma * (x); s4, calculating a value [epsilon] r of an error-based stop criterion based on [mu] * (x) and [sigma] * (x); s5, if epsilon r is smaller than or equal to 0.01, executing the step S6; otherwise, selecting a new training sample # imgabs0 # from the S according to an error reduction learning function, calculating the response of the new training sample # imgabs0 #, adding the new training sample # imgabs0 # into the training set, and returning to the step S2; and S6, calculating a failure probability and a variable coefficient. The method solves the problems that an existing method is insufficient in calculation precision and low in calculation efficiency under multiple failure modes.

In a non-volatile memory device, the amount of delay along an I / O path for transferring received data to a sense amplifier and the amount of delay along a clock path to the sense amplifier can differ. To compensate for this differential delay, a replica of each of the I / O path and the clock path can be included on the memory device. The replica circuits can be used to determine this differential delay between the two paths. To compensate for this difference, an additional delay can be introduced into one or both of the paths, such as in the I / O path, so that any difference in propagations times can be reduced.

The invention discloses a target data selection method based on an approximate Shapley value, electronic equipment and a medium. The target data selection method comprises the steps of obtaining a ratio of a source data set to a target utility, and calculating a total utility value and a target utility value. And screening to obtain an extended sample and a feature set. And sample optimization: carrying out Monte Carlo sampling on the expanded sample set to generate random arrangement, forming a plurality of third subsets and calculating utility values of the third subsets. And after a sample Shapley value is calculated, selecting a subset with the minimum error to carry out iterative sample replacement. And if the error reduction degree is smaller than a second threshold value, re-sampling, otherwise, taking a subset of which the error is smaller than a third threshold value as an initial candidate. And if the final error still exceeds the tolerance threshold, turning to feature optimization. And feature optimization: the process is similar to sample optimization, and feature iteration and replacement are guided through a Shapley value. And alternately optimizing the sample subsets and the feature subsets until the utility value error of the combined candidate set (the candidate sample subsets + the candidate feature subsets) is smaller than an error tolerance threshold.

Examples described herein utilize multi-layer neural networks, such as multi-layer recurrent neural networks to estimate an error-reduced version of encoded data based on a retrieved version of encoded data (e.g., data encoded using one or more encoding techniques) from a memory. The neural networks and / or recurrent neural networks may have nonlinear mapping and distributed processing capabilities which may be advantageous in many systems employing a neural network or recurrent neural network to estimate an error-reduced version of encoded data for an error correction coding (ECC) decoder, e.g., to facilitate decoding of the error-reduced version of encoded data at the decoder. In this manner, neural networks or recurrent neural networks described herein may be used to improve or facilitate aspects of decoding at ECC decoders, e.g., by reducing errors present in encoded data due to storage or transmission.

The invention discloses an inquiry semantic analysis method and device, and belongs to the technical field of computers. The query semantic analysis method comprises the following steps: performing feature extraction on a target statement to obtain a first feature; based on the first feature and a last second feature, obtaining a current slot filling result; performing feature extraction on the current slot filling result to obtain a current second feature; and obtaining an intention recognition result of the target statement based on the current second feature. According to the query semantic analysis method and device, the result of the slot filling task is reversely transmitted to the intention recognition task, the result of the intention recognition task is reversely fed back to the slot filling task, bidirectional information interaction is achieved, the slot filling task and the intention recognition task can be mutually enhanced, and the query semantic analysis efficiency is improved. The risk of error propagation can be reduced, the accuracy of entity recognition, slot filling and intention recognition can be improved, and the overall accuracy of semantic analysis can be improved.

The invention describes a method (100) for reducing error in the determination of an observable of a quantum system in a quantum state on a noisy quantum computer platform, comprising the following steps: • dividing (101) an observable operator into series of base operators; • providing (102) calibration operators; • measuring (103) the calibration operators, comprising determining (104) noisy measurement results (1040) and gate-error-reduced measurement results (1050) of the calibration operators; • ascertaining (106) a calibration function (1060) for mapping the noisy measurement results (1040) to the gate-error-reduced measurement results (1050) of the calibration operators; • determining (107) noisy measurement results (1070) of the series of base operators; • determining (108) error-reduced measurement results (1080) of the series of base operators by mapping (109) the noisy measurement results (1070) by means of the calibration function (1060); • determining (110) the observable by combining the error-reduced measurement results (1080) of the series of base operators.

The application provides a work order processing method and device, the method comprising: obtaining work order data; extracting fault information from the work order data to obtain the fault information; matching the fault information with work order instances in a previously established fault feature database to obtain matched work order instances; wherein the fault feature database is established based on work order instances and corresponding processing strategies; and calling the corresponding processing strategies and executing according to the matched work order instances. The application extracts fault information from the obtained work order data to clearly define the core content of the fault, remove irrelevant information interference, improve the accuracy and efficiency of subsequent processing, matches the extracted fault information with work order instances in the fault feature database to quickly locate the verified processing strategy corresponding to the matched work order instances, improves the accuracy of work order processing, reduces the possibility of incorrect processing, reduces the time and cost of starting from scratch to analyze and solve problems, and improves the speed and success rate of fault processing.

The invention provides a power transmission and transformation project cost control line measurement and calculation generation method, and relates to the technical field of power transmission and transformation project cost control. Step 2, feature extraction and normalization; 3, applying a normal distribution algorithm; 4, applying a normal distribution algorithm; and 5, generating a result. Through a normal distribution algorithm, a big data structured data mode, data cleaning and feature extraction, an efficient and scientific cost control line generation method is effectively provided for the power transmission and transformation project of the Guangzhou bureau, automatic generation, cost management optimization, accuracy and scientificity improvement, manual error reduction and project implementation guarantee are realized, and the method has significant application value.

The invention relates to the technical field of community management, in particular to a community management method and system based on artificial intelligence. The method comprises the following steps: obtaining an owner question content; analyzing the question content of the owner, and determining a question item; based on the question item, utilizing artificial intelligence to obtain item data; summarizing project bills according to the project data; and checking the project bill to obtain and display a final project bill. The problems of delay and fragmentation of manual association of data sources are eliminated, the integrity and consistency of project data are ensured, data missing or conflicts are reduced, and traceability is enhanced. The problems of error proneness and static account division depending on manual summarization are avoided, the accuracy and dynamic adaptability of bills are improved, and the error rate is reduced. The project bill is checked, and the final project bill is obtained and displayed, so that the problems of omission and low efficiency of manual checking are avoided, the accuracy and the displayability of the final project bill are ensured, real-time and safe publicity is realized, the transparency is improved, and the owner trust is enhanced.

Methods and systems are provided for performance optimization of object grouping patterns in storage devices. The methods include receiving object IOs for a target device; grouping the object IOs using a first plurality of input parameters; associating tracking parameters with the first plurality of input parameters and performance parameters corresponding to the first plurality of input parameters; storing first data entries comprising the tracking parameters, the first plurality of input parameters, and the performance parameters in a database; extracting a plurality of data entries from the database, the plurality of data entries including the first data entries; training a training model using one or more of the plurality of data entries; cross-validating the training model to determine a degree of error reduction of the training model; performing a model check to compare the training model to an inference model; and updating the inference model based on the model check.

The embodiment of the invention provides a small sample anti-fraud classification method and device, a medium and a product, and relates to the technical field of artificial intelligence. The method comprises the steps of constructing intra-class semantic features and inter-class difference features based on a preset task framework of small sample learning; determining an anti-fraud classification network model according to the intra-class semantic features and the inter-class difference features; and according to the network model, performing anti-fraud classification identification on a to-be-classified text. According to the scheme, the preset task framework based on small sample learning does not need large-scale labeling, when a category is newly added, only a small number of samples need to be supplemented according to the framework to be fused into an existing model, an overall data system does not need to be reconstructed, the expansibility is greatly improved, through intra-class and inter-class feature learning, the distinction degree of similar fraud classes is enhanced, wrong classification is reduced, and the classification efficiency is improved. The classification accuracy is improved, general fraud features can be migrated to a new scene, and the generalization ability of the new fraud scene is improved.

The present invention provides a robot control method and system based on a large language model. This system employs a dual large language model evaluation mechanism. The large language model of the code generation and modification module uses Claude 3.5sonnet to generate control code, which is then evaluated and optimized using OpenAI's GPT-4O evaluation model. This reduces the risk of difficulty identifying problems during evaluation due to the homogeneity of the same module, thereby improving the quality, safety, and security of the generated code. In addition to the dual large language model evaluation mechanism, the system incorporates safety prompts into the prompts of the generated model and provides generation template examples. This restricts operation within a specific framework, reduces the likelihood of errors, and further ensures the standardization and security of the generated code. Furthermore, a control verification module constructs a simple simulation environment after the control code is generated, enabling early screening of the generated code for syntax errors and logical vulnerabilities, further reducing potential safety hazards.

The invention provides an execution scene adjustment method based on an intelligent rollback mechanism. The execution scene adjustment method comprises the steps of obtaining arithmetic unit information, generating and encrypting revision information, recognizing a target execution scene based on a standard matching degree, and verifying the correctness of the revision information. And by comparing the security level of the architecture, the revision information is ensured to adapt to the execution environment. The verification process comprises the step of comparing a revision record with a master control record so as to ensure consistency and safety. According to the method, an intelligent rollback mechanism is utilized, abnormity is automatically detected, the version is recovered to the previous version under the triggering condition, meanwhile, revision information is optimized in combination with machine learning, updating reliability and adaptability are improved, and the risk of wrong updating is reduced.

The invention provides a tool bar strain measurement device and a machining error reduction method thereof, and relates to the technical field of numerical control lathe machining. The tool bar strain measuring device comprises a tool bar, a sensor mounting groove is formed in the tool bar, a strain sensor is fixed in the sensor mounting groove, a signal transmission channel is arranged at the tail of the tool bar and used for being connected with the strain sensor and a data acquisition device, and the data acquisition device is used for collecting and processing strain signals in real time. Tool bar strain data of the strain sensor are acquired in real time through the data acquisition device, whether machining errors exist or not is quickly judged, and cutting parameters are dynamically adjusted. Machining errors caused by cutting force can be remarkably reduced, universality is high, installation is easy and convenient, the method is suitable for various working conditions, and the machining precision of the numerical control lathe can be effectively improved.

The invention discloses a field expert-based machine question and answer large model illusion question mitigation method, which comprises the following steps: S1, expert field identification: generating a declaration basis for initial reasoning and judgment through a large language model, and identifying experts involved in declaration through a field expert large language model; s2, obtaining a final evidence, identifying a key entity in the statement through a domain expert big language model, taking the key entity as an external evidence for query and retrieval, and selecting an external evidence useful for analysis of the statement as the final evidence; and S3, verifying correctness, comparing the declaration with the final evidence through a domain expert big language model, analyzing the correctness of the declaration, and giving a declaration modification suggestion, so that generation of wrong reasoning is reduced, and the reasoning accuracy is improved.

The invention discloses a large model context error management method and system based on regression testing, belongs to the technical field of artificial intelligence, and aims to solve the technical problems of repeated errors, lack of error learning ability and imperfect error prevention mechanism existing in a large model at present. Comprising the following steps: constructing an error history library based on a triple data structure corresponding to historical error data; performing similarity analysis on the current user input and historical user input in an error historical library; generating a test case based on a historical user input similar to the current user input; optimizing the current user input to obtain optimized user input; and establishing an error reduction rate evaluation index, comparing error occurrence rates before and after user input optimization through an A / B test framework, and adaptively optimizing a similarity judgment standard, a weight parameter and an error avoidance instruction template based on a feedback result.

A robot system includes: a robot having an arm; and control circuitry configured to: store factor information representing an error factor of a motion of the robot, wherein the error factor is a mechanical characteristic of the robot that causes a positional error of an extremity of the arm and wherein the factor information has been predetermined based on a comparison between a programmed motion pattern and an actual motion of the robot that is operated according to the programmed motion pattern; calculate, based on a taught position of the robot and the factor information, a positional error of the extremity that would occur during an expected motion toward the taught position; and control, based on the taught position and the positional error, the robot to move the extremity toward the taught position with a positional adjustment of the robot to reduce the positional error.