6results about How to "Address privacy concerns" patented technology

A feature library construction method, the method comprising: obtaining a first file sample set, the first file sample set comprising a plurality of file samples and a cleaning level of each file sample; extracting a feature word that can affect the cleaning level of each file sample from each file sample in the first file sample set; filtering the extracted feature words to obtain a plurality of effective feature words; setting a weight of each effective feature word, and constructing a feature library for recommending the cleaning level of a file according to each effective feature word and the weight of each effective feature word.

This disclosure provides a natural interaction method and system for AR glasses QR code payment scenarios, applied in the field of AR technology. The method includes: responding to a user's activation command, the AR glasses acquire an environmental image of the area to be collected; wherein the user's hand is placed in the area to be collected with a specific gesture; based on the specific gesture, a lightweight image recognition model is used to identify whether the user in the environmental image has payment intent; if so, the corresponding payment code is identified; it is determined whether the identified payment code conforms to the static payment protocol specification; if so, a voice monitoring module is activated to acquire environmental sound, and the amount information is obtained based on the environmental sound; payment is performed after the user confirms the amount information. This disclosure integrates the payment process with natural offline interpersonal interaction, avoids accidental triggering through precise pointing selection, and embeds the payment amount input into the necessary dialogue between the user and the merchant, eliminating the awkwardness of the user publicly reciting the amount; finally, authorization is completed through physical triggering.

The invention discloses a hidden income recognition method and device based on agent evolution and a medium, and relates to the technical field of hidden income recognition. The method comprises the steps that original data are acquired and encrypted through a plurality of agents deployed in a local data source, an encrypted state association graph is obtained, and the original data comprise information flow, capital flow, logistics and bill flow; decrypting the encrypted state association graph, and distributing the encrypted state association graph to a federated network for independent analysis to obtain an abnormal clue fragment; encoding the abnormal clue fragment into a feature vector, and mapping the feature vector into a symbol predicate; loading a predefined differentiable tax rule knowledge base, performing micro reasoning on symbol predicates to generate risk judgment confidence, and backtracking rules and predicates activated in the micro reasoning process to generate a structured evidence chain. According to the method, correlation analysis of original sensitive data of taxation parties, banks, logistics parties and the like can be completed without plaintext collection, and privacy and compliance bottlenecks of data fusion are fundamentally solved.

This utility model discloses a car center console storage box, relating to the field of automotive parts technology. It includes a storage base plate, a cover plate snapped onto the top of the base plate, and two symmetrical storage side plates snapped onto both ends of the base plate. Two symmetrical sliding grooves are formed on the inner side of each side plate. In this utility model, the car center console storage box consists of a storage base plate, storage side plates, a silicone movable plate, and a cover plate. Through the cooperation between the storage side plates and the silicone movable plate, and the design of the movable silicone plate, the silicone movable plate and the storage side plates are fixed by a hidden magnetic structure, facilitating easy access for the driver or passenger. This not only facilitates user operation but also enhances the safety and practicality of the storage box. Simultaneously, it completely covers items on the cover plate, effectively solving the problem of exposed items affecting aesthetics and privacy, while keeping the cover plate and items clean. During vehicle operation, it prevents items from slipping and being damaged, ensuring the safety of the items.

The application discloses a kind of virtual power plant user regulation capacity prediction method and system based on edge computing and privacy protection in the technical field of power system automation and demand side response.It is method that includes: according to user side local data, each user load in target scheduling period is predicted using lightweight load prediction model, and the predicted load sequence of each user is obtained;According to the predicted load sequence of each user, the adjustable resource constraint signed by user and the local preset operating constraint condition, the adjustable power interval of each user in target scheduling period is calculated;The adjustable power interval of each user in target scheduling period is aggregated, and the overall adjustable capacity interval is obtained;Market bidding strategy or grid response instruction is generated based on the overall adjustable capacity interval.The application introduces adaptive feedback and privacy-enhanced federated learning mechanism under the premise of ensuring data privacy, to support virtual power plant to participate in power spot market transaction and grid auxiliary service scheduling safely and efficiently.

This disclosure relates to the field of model training technology, and in particular to a model training method and system. It includes: constructing a cold-start dataset for the security field, which contains multimodal security scenario data, target scenario questions, and corresponding text responses; secondly, adjusting the parameters of a basic multimodal model based on this cold-start dataset to obtain an adjusted model with preliminary security recognition capabilities; then, acquiring difficult samples, including failure samples, prediction bias samples, and adversarial generation samples selected from the test set; finally, strengthening the adjusted model based on these difficult samples, ultimately obtaining a high-performance, robust multimodal model for the security field. This solves the technical problem of poor model performance in the security field due to insufficient training data.