626 results about "Confidence value" patented technology

Systems and methods improve security of vehicle-transfer processing using model-driven fraud flagging with cryptographic provenance. A transfer request message including a vehicle identifier is received. Historical event records for the vehicle identifier are retrieved from an append-only, hash-linked event log. A feature vector is generated from the historical event records and one or more external authoritative data sources. A trained fraud detection model is executed to produce a fraud score and, in some embodiments, a confidence value. When a fraud-flagging criterion is satisfied, a fraud-flag artifact is generated including a model identifier and version, a feature schema identifier, and a hash of at least a portion of the feature vector, and the artifact is digitally signed. Based on the signed artifact, the system selectively blocks, delays, or routes the transfer request to remediation, preventing commitment of a corresponding transfer record to the event log until a remediation condition is satisfied.

In various examples, data collection vehicles or machines may be equipped with one or more LiDAR sensors (and / or other sensors), and the LiDAR sensor(s) may be used to collect frames of LiDAR data representing various real-world conditions. The LiDAR data may be processed using one or more deep neural networks (DNNs) such as a transformer neural network to generate auto-labels representing detected dynamic obstacles of any designated class. Tracking may be applied to generate object tracks (tracklines), estimate velocity, and / or handle occlusions. In some embodiments, the object tracks may be refined based on geometry and / or confidence to improve their accuracy. In some embodiments, the auto-labels are classified to generate an estimated representation of quality, and auto-labels with at least a threshold quality score may be skipped during human labeling. As such, auto-label quality scores may be used to accelerate human validation of auto-labeled scenes by skipping high quality auto-labels.

This application provides a business data detection method, system, medium, device, and program product, relating to the field of network data security. The method includes: outputting security detection results based on abnormal business data; if the security detection results do not meet a confidence threshold, determining a data object to be invoked based on the abnormal business data; sending the data object to be invoked to a security detection agent, so that the security detection agent can use the data invocation tool corresponding to the data object to obtain security-related data related to the abnormal business data; the security-related data includes at least one of time-related data, content-related data, and identity-related data; and outputting a security detection result that meets the confidence threshold based on the abnormal business data and the security-related data. This application can improve the detection accuracy of business data.

The invention discloses a medical image treatment method and system for multi-source heterogeneous data, and relates to the field of medical image treatment, and the method comprises the steps: firstly, respectively extracting an image embedding vector and a text embedding vector from original medical image data and a description text thereof through a deep learning model; then, the vectors of the two different modes are fused, and a unified semantic embedding vector is formed; based on the unified vector, through semantic similarity calculation with a standard term library, candidate mapping can be automatically generated, and dependence on rigid artificial rules is eliminated. More importantly, a closed-loop mechanism of manual auditing-feedback learning is introduced in the scheme, high-confidence mapping is adopted automatically, low-confidence mapping is audited by experts, and an auditing result is absorbed into a mapping knowledge base, so that the system has continuous learning and self-evolution capabilities, and a semantic gap of cross-mechanism data can be eliminated more intelligently and more accurately.

An evaluation system, method and program product based on a large language model relate to the technical field of artificial intelligence. The system obtains an evaluation object and an evaluation task configuration, constructs an evaluation structure graph containing theme segments, argument segments, basis segments and reference segments and their support, reference and conflict relations, filters key semantic units, and constructs a coupling representation between key semantic units, evaluation dimensions, dimension rules, candidate evidence segments, evidence positions and unit-evidence relationship markers; input the coupling representation into a large language model to generate an initial evaluation result; construct a statement disturbance set around the key semantic units that keeps the evaluation theme and core facts unchanged, obtain consistency indicators through re-evaluation, and calculate joint confidence by combining evidence coverage, evidence conflict and evidence chain drift; and perform evidence weighting and local calibration on low-confidence target evaluation dimensions. The scheme can improve the accuracy, stability, explainability and traceability of the evaluation result.

This application relates to the field of fintech and discloses a dynamic hybrid parsing and deterministic verification system and method for fuzzy bond instructions. The system constructs structured data through an instruction preprocessing module; a routing decision module calculates complexity scores based on structure, semantics, and business characteristics, and generates path selection instructions using a resource-aware dual-threshold strategy; a hybrid parsing module responds to instructions, employing a rule engine to handle low-complexity tasks and a large-scale model with enhanced retrieval to handle highly ambiguous tasks; a fallback verification module performs confidence-weighted integration and logical consistency checks, and repairs anomalies through issuer set operations and edit distance matching; a mapping generation module constructs standard query statements using an abstract syntax tree; and a closed-loop optimization module dynamically updates parameters based on user interaction feedback. This invention effectively balances computational efficiency and parsing accuracy, eliminates model illusion risks, and achieves adaptive evolution of the system.

This invention discloses a method for the collaborative optimization of turbine governor parameters and hydraulic system parameters. The method involves synchronously collecting multi-source operating data from a hydropower station using a sensor network to construct a hydraulic-governing coupled state-space model. Based on the UKF unscented Kalman filter algorithm, the method identifies the water flow inertial time constant, water hammer wave velocity, and turbine integrated transfer coefficient online, and evaluates the reliability of the identification results using a parameter identification confidence index. In a joint high-dimensional space of the control parameter domain and the hydraulic parameter domain, the method uses regulation quality, hydraulic safety, and grid stability as three objective optimization functions to solve for the Pareto front candidate set, and selects robust parameter schemes using a robust stability margin index. A three-level differentiated push strategy is implemented based on the parameter change amplitude, and real-time monitoring and an automatic anomaly backoff mechanism are implemented during the gradual push process. This invention achieves full-cycle collaborative optimization and safe tuning of hydraulic system parameters and governor control parameters.

A system receives staging test results associated with a software product from a staging testing unit. The staging test results are obtained by testing the software product with a test set that imitates a real production environment. The test set includes associations of a set of customer-facing service (CFS) features and a set of network-facing service (NFS) features, and the staging test results include identified anomalies arising from the associations of the set of CFS features and the set of NFS features. The system processes, using an artificial intelligence model, the staging test results to produce a confidence value indicating whether the identified anomalies have a likelihood of causing a failure of the software product when deployed in the real production environment, and determines, using the confidence value, whether to deploy the software product in the real production environment.

This invention discloses a method, system, device, and medium for generating and presenting a large-scale modular fact-checking response. The method includes: acquiring user input instructions and decomposing them into information fragments to be checked; retrieving and aggregating evidence sets from a pre-set knowledge base or online information sources; determining the consistency of the information fragments based on the evidence, and outputting the verification conclusion and confidence level; determining the risk level and generating control parameters according to a pre-set risk assessment strategy; and using the control parameters to control or adjust the generation and / or perceptual presentation of the response content. The control or adjustment methods include constraints during generation and post-processing corrections to suppress deterministic assertions and output uncertainty identifiers and evidence summaries when evidence is insufficient or conflicting. This solution improves factual consistency and traceability through decoupling of responsibilities and an auditable evidence chain, and reduces the risk of illusion caused by consistency overfitting. The risk level can be further mapped to decoding parameters of the generation module (such as temperature, top-p, and logit bias masks) or modulation parameters of the perceptual presentation. The response data can be provided to the client via a network interface, allowing the client to control the output presentation.

The present application belongs to the technical field of industrial internet security, and relates to an industrial internet vulnerability library establishment method, which solves the problems of low construction quality, weak traceability, insufficient continuous evolution ability and difficulty in supporting deep security application of the existing vulnerability library. The present application obtains industrial internet multi-source vulnerability data and external feedback data, obtains extraction results and initial confidence based on multi-source vulnerability data by adopting rule extraction and remote supervision deep learning joint extraction, divides the certainty knowledge base and the to-be-reviewed queue according to the double threshold after the confidence is optimized by the graph convolution network, analyzes the external feedback data of the samples in the to-be-reviewed queue as the instant reward signal, optimizes the sampling strategy and updates the model through deep reinforcement learning, extracts triples from the certainty knowledge base to build a knowledge graph and generate a version hash chain regularly, and obtains a structured vulnerability knowledge base containing a hash chain and confidence evaluation. The present application realizes high-precision construction and dynamic optimization of the vulnerability library.

This invention relates to the field of distributed computing technology, and in particular to a unified registration and scheduling system for AI agent capabilities in edge-cloud collaboration, comprising: a dynamic behavior mapping module for capturing client interaction behavior features and mapping them to behavior control topology, performing path adjustment when deviations occur, and registering the control behavior to a unified MCP capability center; a semantic feature compression module for dimensionality reduction and compression of the original multimodal data stream to generate low-dimensional semantic feature stubs; an asynchronous state alignment module for converting feature stubs into asynchronous stateless message streams and reconstructing context mirror data on the cloud side; an inference stream scheduling module for generating inference stream scheduling strategies based on multi-dimensional states; a thought chain flow control module for dynamically truncating the streaming output of the large model according to the strategy and extracting core control segments and confidence levels; and a residual collaborative execution module for sending the core control segments to the edge for residual logic completion, driving the topology to execute control behaviors.

This invention provides a method and apparatus for 3D target detection in autonomous driving, relating to the field of autonomous driving technology. The method includes: acquiring point cloud data collected by LiDAR; for each point in the point cloud data, defining a neighborhood of a set radius centered on the current point, counting the number of points within the neighborhood, and normalizing the number of points within the neighborhood to obtain a normalized value; determining the density feature of the current point based on the negative of the normalized value; extracting semantic features of each point in the point cloud data through multiple SDSA layers, predicting the foreground confidence of each point based on the semantic features, and performing point sampling by combining the density features and the foreground confidence to generate sampling points; aggregating the features of each sampling point to generate aggregated features for each sampling point, and predicting 3D target bounding boxes based on the sampling points and their aggregated features. This improves the sampling bias problem caused by the unevenness of the point cloud data.

This invention discloses a target detection method based on variable threshold and matrix calculation, relating to the fields of computer vision and artificial intelligence application technology; including: Step 1: For detection images containing small targets, preprocess all detection boxes output by the detection model; Step 2: Preset the IoU value as the density threshold. If and only if the IoU between two detection boxes is greater than or equal to the density threshold, it is determined that the two detection boxes constitute a spatial proximity relationship. Based on this proximity relationship, for each detection box, count the total number of neighboring detection boxes that meet the density threshold condition, and define the statistical value as the local spatial density value of the detection box; Step 3: Map the local spatial density value one-to-one to the dynamic confidence threshold corresponding to the detection box, and expand the N-dimensional vector of the dynamic confidence threshold of N detection boxes into a threshold matrix T with dimension N×N; Step 4: Compare the IoU matrix I with the threshold matrix T position by position; Step 5: Compare the summation vector S with the confidence vector C element by element: For each detection box i, only the detection boxes that satisfy the condition S[i]≤C[i] are retained as the output result for subsequent target detection processes.

The application discloses a gearbox fault diagnosis method based on adaptive decomposition and transfer learning, comprising: collecting multi-condition vibration signals as original input; adopting an exponential coupling resonance search-Fourier adaptive modal decomposition method to adaptively decompose the signals, accurately decoupling multi-component signals and suppressing modal aliasing; extracting multi-domain features and utilizing weighted multi-objective probability principal component analysis dimension reduction to obtain a sensitive low-dimensional feature set; constructing a sparse excitation residual network with a sparse constraint channel attention mechanism as a feature extractor and pre-training; adopting a confidence-aware incremental open set transfer learning strategy to fine-tune the high layer of the network, realizing known fault diagnosis and open set identification of unknown fault samples. The method effectively improves the fault feature decoupling capability and the generalization performance of the diagnosis model, and is suitable for intelligent fault diagnosis and state monitoring of the gearbox under multiple conditions.

This invention discloses a method for quantifying anxiety and depression comorbidities using partial label learning based on brain imaging, belonging to the fields of medical image processing and artificial intelligence. The method is based on a dataset constructed using brain imaging measurements. It selects representative typical anxiety and depression samples with distinctive brain imaging features from clinically diagnosed groups and constructs a sparse nearest neighbor graph by comprehensively considering the relationships between each sample and its nearest ordinary and typical neighbor samples. Partial label learning modeling and optimization are performed through an iterative label propagation mechanism, ultimately outputting label confidence matrices for clinical anxiety patients, depression patients, and patients with comorbid anxiety and depression. This enables a quantitative assessment of the anxiety and depression levels in patients with comorbid anxiety and depression and completes the classification of comorbidity subtypes. This invention improves the accuracy and objectivity of anxiety and depression comorbidity assessment, providing auxiliary support for the clinical diagnosis, pathological mechanism research, and intervention target screening of anxiety and depression comorbidities.

Systems, methods, non-transitory computer-readable media to perform operations associated with the storage medium. One system includes a storage medium and an encoding / decoding (ED) system to perform operations associated with the storage medium, the ED system being configured to process a set of log-likelihood ratios (LLRs) and a syndrome vector to obtain a set of confidence values for each bit of a codeword, estimate an error vector based on selecting one or more bit locations with confidence values from the set of confidence values above threshold value and applying hard decision decoding to the selected one or more bit locations, calculate a sum LLR score for the estimated error vector, and output a decoded codeword based on the estimated error vector and the sum LLR score.

This invention discloses a method and system for extracting domain knowledge graphs based on the collaboration of large and small models. First, central entities are extracted and filtered. Then, the text is segmented to generate associated summaries, forming enhanced semantic blocks, which are then vectorized to construct a knowledge base. Next, a large model and a domain segmenter are used in parallel to extract candidate entities, which are then filtered by type discrimination and confidence to generate entities with contextual descriptions. Then, based on entity pairs, the supporting text is retrieved, and usability, relation type, illusion, strength, and direction detection are performed sequentially. If successful, relation triples are generated. Finally, a star graph and similarity support material are constructed for the entities. Through two-stage semantic retrieval and large model judgment, the fusion of entities with the same name is achieved, and the association relationships are updated. This invention can automatically complete the process of constructing a high-quality knowledge graph from text under the condition of a single-machine, relatively parameter-based model.

This invention provides a method for security responsibility assessment and closed-loop traceability in smart governance scenarios, comprising: acquiring multi-source heterogeneous data in the smart governance scenario; generating a corresponding event identifier when an abnormal event is detected in the multi-source heterogeneous data; extracting a structured evidence package within a target time window and target spatial range based on the event identifier; extracting responsibility factors from the structured evidence package; inputting the responsibility factors into a responsibility assessment model and a responsibility scoring function to obtain the responsibility score, responsibility level, and responsibility confidence level corresponding to the abnormal event, and determining at least one of the direct responsible entity, related responsible entity, and management responsible entity based on the responsibility correlation; and automatically generating corresponding handling instructions based on the responsibility level and the category of responsible entity.

The application discloses a kind of legal document quality confrontation test method based on logic game, it is related to legal artificial intelligence technical field, including, collection legal document full-text data, and extract fact triple, evidence citation fragment, law article number and conclusion claim, generate structured argumentation set;Based on the confidence attenuation sequence of multiple rounds of game, in combination with the vulnerability propagation coefficient corresponding to different types of edges in directed heterogenous graph, carry out path vulnerability contribution value calculation, and generate legal document global vulnerability index by stratified accumulation;Based on logic breakpoint list, contradiction conflict list and legal document global vulnerability index, comprehensive evaluation legal document's logic integrity, argumentation consistency and conclusion robustness, generate structured check report.The application provides a complete and feasible implementation path for the automatic check of legal documents combined with natural language processing.

The application provides a kind of based on machine learning's tender document analysis processing system and method, through multi-channel receiving clarification request submitted by tenderer and uniform collection;Clarification request is text preprocessed, text features are extracted in combination with natural language processing (NLP) and machine learning technology, key problems and problem types are identified;Similar cases in historical database are retrieved based on problem features, and a candidate set of reference answers is generated;The optimal reference answer is filtered and pushed through confidence calculation, and the interaction details are recorded to form structured knowledge points;Model parameters and processing procedures are iteratively optimized based on knowledge points. Using the scheme of the application, automated and efficient processing of clarification requests can be achieved, reducing manual intervention, avoiding message omission, improving response speed and professionalism, and forming a reusable knowledge system.

The application provides a hyperspectral anomaly detection method and device based on confidence guidance fusion. By introducing a confidence guidance adaptive fusion strategy, effective cooperation of local contrast information and global structure information is realized without complex iterative optimization or large-scale training data support, problems such as unstable anomaly response, high false alarm rate and insufficient feature fusion capacity of existing methods in a complex background are solved, efficient and stable detection of a hyperspectral anomaly target is realized, and the method has good engineering application value.

This invention provides a phase recognition enhancement method and system based on frequency domain analysis and contrastive learning. The method includes: acquiring the user load time series of a user to be identified and the phase load time series of each phase within the same transformer area as the user to be identified; performing phase allocation on the user load time series and each phase load time series based on a dynamic programming algorithm to obtain candidate phases corresponding to the user to be identified and their phase position confidence; determining whether a phase to which the user to be identified belongs exists based on the confidence; if not, extracting frequency domain features from the user load time series and each phase load time series, calculating the correlation between the frequency domain features of the user to be identified and the frequency domain features of each phase, and determining whether a phase to which the user to be identified exists; if not, inputting the user load time series of the user to be identified into a phase recognition model to obtain the phase to which the user to be identified belongs. This invention improves the accuracy and applicability of phase recognition.

The application relates to the technical field of operation and maintenance cross-domain access, in particular to a cross-domain access control method of a distributed operation and maintenance audit system, which comprises the following steps: obtaining a high-risk degree according to the high-risk confidence of an instruction and a global consequence performance value, and then obtaining an asset sensitivity degree; obtaining a real-time abnormality index according to the current high-risk prominence degree of a security domain and the mean value of the high-risk degrees of various instructions within a first preset time; obtaining a real-time sensitivity degree based on the real-time abnormality index and the asset sensitivity degree; obtaining a cross-domain sensitive change degree based on the real-time sensitivity degrees corresponding to the initiation domain and the target domain of an instruction; obtaining a cross-domain risk degree based on the cross-domain sensitive change degree, the maximum value of the real-time sensitivity degrees in the security domains involved by an access link corresponding to the instruction and the historical maximum high-risk degree of the instruction; and executing a corresponding level of release strategy on the instruction according to the cross-domain risk degree. The application can significantly enhance the overall security and active defense capability of cross-domain access.

This application discloses a search error correction optimization method, apparatus, device, and computer program product. The method includes: acquiring user interaction behavior logs, including search behavior logs and click behavior logs; preprocessing the interaction behavior logs; performing user behavior analysis on the preprocessed interaction behavior data to obtain user behavior analysis results; making decisions on the preprocessed interaction behavior data based on the user behavior analysis results using a pre-set confidence decision threshold to obtain decision results; and updating the search error correction terminology in a closed loop based on the decision results. This application constructs a search error correction closed-loop optimization scheme based on user behavior feedback, enabling the error correction system to have self-evolution capabilities; it upgrades from "algorithm-driven" to "user intent-driven" based on massive amounts of real user behavior data, more accurately reflecting user needs; and it improves decision reliability through scientific quantitative decision-making using a confidence decision model.

This invention provides a method and system for incremental retrieval index update of heterogeneous data, relating to the field of data processing technology. The method includes the following steps: batch transferring newly added information representation values ​​from the real-time cache layer to the recently active layer and performing local index optimization; analyzing the information representation values ​​in the recently active layer and the information representation values ​​in the historically stable layer and establishing jump edges; identifying invalid pointers and duplicate information fragments in the index for quickly finding similar information, and deleting invalid pointers and merging duplicate information fragments. This invention aims to solve the problems of traditional incremental update methods failing to effectively integrate different types of data when intelligent question-answering systems process massive, heterogeneous data, leading to limited retrieval efficiency and accuracy, as well as inconsistencies in internal index logic, invalid pointers, and duplicate information fragments, effectively improving system response speed and query result confidence.

The present application relates to the technical field of data processing, and specifically discloses an electronic insurance policy intelligent verification and repair method, system, device and medium. The present application extracts structured insurance policy data through multi-modal analysis, performs hierarchical processing in combination with a large model semantic verification and confidence evaluation model, respectively executes automatic repair or generates a manual processing task carrying repair guide information according to different confidence levels, and iteratively optimizes the large model and the confidence evaluation model based on repair process data and results, thereby solving the problems of low artificial sampling efficiency, high failure and leakage rate of rule templates facing non-standard insurance policies, uncontrollable output of general large models, and lack of confidence calibration and closed-loop repair capabilities, and achieving dual improvement of electronic insurance policy verification accuracy and automation processing efficiency.

The application discloses a generative AI industrial design topology closed-loop optimization method and system, and relates to the technical field of computer-aided design, comprising: obtaining an initial three-dimensional geometric representation of an industrial product and design constraints, and constructing a geometric parameter offset field; in iterative deformation, calculating discrete curvature, deformation energy, time sequence mutation and topology anomaly index, triggering and positioning a topology sensitive area; in the area, performing edge collapse, edge split, local re-meshing, Loop subdivision, constraint Laplace smoothing or SDF local reconstruction, and verifying and repairing the results by Euler characteristic number, genus, boundary ring, connected component and topology anomaly; calling a physically consistent proxy model to predict performance and evaluate confidence, triggering finite element sampling when confidence or safety margin is insufficient, error risk and topology sensitive area overlap, updating the proxy model by using finite element true value increment, and finally completing layered optimization and complete finite element verification.

The present application relates to a kind of methods for realizing network content risk analysis based on key points driving, comprising the following steps: depth semantic analysis is carried out to original text, and risk key points are identified and extracted;According to the preset confidence threshold filtering low reliability key points, calling big model intelligent screening risk key points;Multi-source evidence online retrieval is carried out, and multi-source search results are integrated and risk tendency determination;Multi-dimensional risk research and judgment is carried out.The method for realizing network content risk analysis based on key points driving, device, processor and computer readable storage medium thereof of the present application are adopted, by reconstructing analysis process with risk key points as core, and comprehensively deciding by depth fusion local semantic knowledge and external real-time information, surpassing existing technology in precision, explainability and flexibility etc.Dimensions, provide solid technical foundation for constructing next-generation intelligent, efficient, reliable network content risk prevention and control system, and have important practical value and innovation.