8 results about "False rejection" patented technology

The invention discloses a self-service door opening method for a store, and relates to the technical field of self-service stores. The system carries out intelligent matching according to a pre-stored user file, calculates a similarity threshold value through an algorithm, judges that a user is a legal user if the matching degree exceeds a preset value, or rejects access; a dynamic combination strategy is adopted to allow an administrator to configure weights and combination modes of different fields according to the security requirements of the store, so that the user experience is optimized and the privacy leakage risk is reduced on the premise of ensuring the security; a background management system flexibly sets a weight value and combinatorial logic of each verification element, the matching efficiency is dynamically optimized based on a preset similarity threshold, if a comprehensive score after element combination reaches a security level score configured by the system, rapid door opening is automatically authorized, otherwise, secondary verification or manual auditing is triggered, and the safety of the system is ensured. The false rejection rate is reduced, the safety protection capability is improved, the verification modes are automatically switched according to regions or time periods, and the door opening speed is high when the operation is successful.

The invention discloses an AI-based dialogue type zero-trust permission policy generation method and system, and the method comprises the steps: receiving user request information which comprises term information; the term ambiguity in the user request is analyzed through the domain knowledge graph, and an intention confidence score is obtained; when the intention confidence score is lower than a preset threshold value, generating an interactive clarification option; determining a user clear intention according to the user clear option answer; performing multi-modal verification on the clear intention of the user; and generating a permission minimization strategy based on a multi-modal verification result. Through enterprise-level knowledge graph analysis term ambiguity, semantic analysis accuracy is improved, closed options are generated at low confidence, user intentions, behavior characteristics of text input and authenticity of equipment sensor data verification intentions are dynamically defined, social worker attacks are prevented, and space-time limited permission strategies are generated in real time based on verification results. And the meanings of the terms in the user request are distinguished, so that the situation of excessive permission granting or wrong rejection is avoided.

According to the biological feature recognition method and device, the storage medium and the computer equipment provided by the invention, the pre-trained feature recognition model is determined, and in the training process of the feature recognition model, the model is trained by introducing the target loss function formed by the diversity promotion loss and the feature recognition loss; the feature space distribution of the model can be optimized. Furthermore, the diversity promotion loss enables the inter-class features in the feature recognition model to be separated from each other and the boundary to be clear by minimizing the similarity of different classes of features and applying constraint punishment of the angle level of the different classes of features, so that the model can learn the features with higher discrimination so as to improve the recognition capability. And finally, the target biological features are input into the feature recognition model, and the model can output a more reliable feature recognition result by virtue of a learned strong discriminative feature space, so that the probability of occurrence of mismatching and false rejection is fundamentally reduced.

This invention discloses a method and system for suppressing false rejection behavior and selectively forgetting harmful semantics in large language models across vertical domains based on feature subspace decoupling, belonging to the field of fine-tuning and alignment technology for large artificial intelligence models. The method includes steps for target definition and data construction, feature subspace localization, key level optimization, orthogonal decoupling weight correction, constrained norm renormalization, selective fine-tuning, false rejection calibration, and closed-loop verification monitoring and document output. The system includes modules for concept discovery and data construction, representation subspace modeling, false rejection localization and calibration, and verifiable evaluation and continuous monitoring. This invention can significantly reduce the false rejection rate and maintain a safety baseline while preserving the model's generalizability, meeting the compliance requirements of generative artificial intelligence services.

Systems and methods for performing testing of a single analyte on a group of multiple clinical diagnostic analyzers, or a single clinical diagnostic analyzer having multiple analytic units, or combinations thereof, are disclosed. A mean and SD for each individual instrument are input to at least one of the instruments along with a QC rule to be used, the probability of false rejection function for the QC rule, and a desired false rejection rate. A group mean and a group SD are calculated to satisfy the desired false rejection rate and QC rule and loaded into each individual instrument for use in testing the single analyte at each individual instrument.

This invention provides a method for constructing high-quality metal material process datasets based on a large language model, relating to the field of process data collection and processing technology. It introduces a chemical composition hashing (HC) mechanism, which transforms complex composition information into unique identifiers through standardized concatenation of a preset element set and MD5 hashing, enabling efficient and accurate cross-file comparison. A three-level duplicate detection system—precise duplicates → similar duplicates → LLM intelligent confirmation—effectively avoids false rejections due to differences in expression while ensuring high recall. A prompting engineering system is constructed based on a locally deployed large language model, integrating domain knowledge templates and strict output format constraints. This achieves complete reconstruction and structured expression of complex process chains without manual intervention, enabling end-to-end extraction of standardized material entries from massive amounts of unstructured documents. Furthermore, semantic-level feature hashing and LLM-assisted judgment mechanisms significantly improve deduplication accuracy and logical consistency.

The application relates to the technical field of network security and artificial intelligence, and specifically discloses a damage vulnerability sample generation method based on self-consistent explanation, which first extracts damage scene features from multi-source situation data, maps the features into a structured thinking chain of syllogistic logical reasoning of a forced large model according to vulnerability principle-triggering condition-damage consequence, generates codes after establishing a complete cause-effect chain from the source constraint model, and avoids statistical probability-driven shallow imitation. Subsequently, the generated code segments and mechanism assertion texts are respectively subjected to static topology reverse deduction and semantic coding, cross self-consistency verification is realized by calculating cosine similarity, and false defect samples are filtered. Further, an attack chain cascading dependence graph and link propagation enhancement scoring are introduced, the cause-effect conduction relationship among samples is brought into evaluation, false rejection of attack chain bridging nodes and false retention of logically contradictory samples are avoided, and finally a high-fidelity vulnerability sample library that can be used for multi-level damage scene testing in an industrial field is generated.

The invention discloses an intelligent traceability drawer lock system for refined anesthesia medicines, and relates to the technical field of medicine storage and taking. The precise anesthesia medicine intelligent tracing drawer lock system comprises a multi-mode biological recognition module, an early warning control module, a block chain tracing module and a dynamic permission execution module, and the multi-mode biological recognition module, the early warning control module and the block chain tracing module are electrically connected with the dynamic permission execution module. Multi-modal identification fuses fingerprint and palm vein dual biological characteristics, the problem that a single fingerprint loses efficacy due to moisture and abrasion is avoided, palm vein non-contact acquisition is more suitable for the operation scene that medical staff wear gloves, the false rejection rate and the false acknowledgement rate are both superior to the industrial standard, it is ensured that identity verification is accurate and efficient, and the edge computing chip and the AI algorithm are combined with multi-sensor data, so that the accuracy and accuracy of identity verification are improved. Potential abnormal behaviors can be pre-judged in advance, a traditional passive response mode is broken, the fine anesthesia medicine management risk is reduced from the source, and the safety protection level is improved.