67 results about "Sequential data" patented technology

The invention discloses an abnormal Transform model-based electricity larceny detection method and device and a storage medium, and relates to the field of electric power system electricity larceny detection, and the abnormal Transform model-based electricity larceny detection method comprises the following steps: capturing complex correlation among different time points in sequence data by using a self-attention mechanism of a Transform model; the characteristics of the normal time point and the abnormal time point are effectively distinguished by learning the relevance difference of each time point, the accuracy of electricity stealing user identification is improved, and the misjudgment rate is reduced. The method comprises the following steps: S1, obtaining user multi-dimensional power consumption time series data, preprocessing the data, and constructing a sliding window sample; s2, constructing an electricity larceny detection model based on an Angle-Attention mechanism, wherein the electricity larceny detection model comprises a prior correlation branch and a sequence correlation branch; s3, quantifying the time dependence deviation of the power consumption behavior by taking the relevance difference as an anomaly judgment standard; s4, training the model by using a minimax optimization strategy, and amplifying the relevance difference between normal electricity utilization and electricity stealing electricity utilization; and S5, performing electricity larceny risk prediction and model performance evaluation through the joint score of the relevance difference and the reconstruction error.

A system and method for robotic teleoperation enable a teleoperated robotic element to perform a sequence of actions based on intention estimation, eliminating the need for continuous human control. The system includes a robotic teleoperation input that receives motion inputs and gaze data from a human operator performing a robotic teleoperation task. A robotic teleoperation feature extractor analyzes and processes the motion inputs and the gaze data into sequential input data. A multi-window model assigns hierarchical prediction windows to the input data, generating windowed sequential data. A hierarchical neural network processes the windowed sequential data to determine low-level action intentions and high-level task intentions, and generates an intention estimation based on the low-level action intentions and the high-level task intentions. A hierarchical dependency model incorporates hierarchical dependent loss to refine the intention estimation.

The invention discloses a multi-source information fused intelligent inspection system, method and equipment with a target guiding function. The system comprises an inspection management platform, a data base station, a wireless routing base station, a quadruped robot, a pose sensor and a multi-mode sensor unit, wherein the inspection management platform, the data base station, the wireless routing base station, the quadruped robot, the pose sensor and the multi-mode sensor unit comprise a laser radar, a visible light camera and an infrared thermal imager; a control processing terminal is arranged on the quadruped robot and is used for acquiring data acquired by the multi-mode sensor unit and the pose sensor, processing the acquired data and outputting a control signal according to external instruction data, so that the quadruped robot moves according to an inspection planning path; the multi-mode sensor unit and the pose sensor are controlled to work; and the inspection management platform is used for analyzing and processing the data acquired and processed by the control processing terminal, and outputting instruction data including an inspection planning path to the control processing terminal. According to the invention, the inspection safety of engineering scenes in dangerous environments such as underground spaces and tailings ponds can be improved.

The invention discloses a wind turbine generator component life probability prediction method and system, and a wind power plant energy control method and system. The prediction method comprises the steps of constructing a state space model of a degradation process of a wind turbine generator component; acquiring real-time state information of the component; inputting the real-time state information into the state space model, evaluating the degradation state of the component by adopting a sequential data assimilation algorithm, and outputting posterior probability distribution of the degradation state; and based on the posterior probability distribution, forward simulating a future degradation path of the component by adopting Monte Carlo simulation until the future degradation path exceeds a preset failure threshold, and obtaining residual life probability distribution of the component. According to the method, the personalized residual life of each unit can be accurately evaluated, and an innovative wind power plant energy management control strategy is created on the basis, so that the absolute maximization of the full-life-cycle economic benefit of the wind power plant in a complex market environment is realized.

The invention provides a self-adaptive cardinality estimation method, system and device and a storage medium, and belongs to the field of database query optimization, and the method comprises the steps: constructing original connection sequence data and an optimal table sequence of a table based on a disclosed benchmark test data set as a data set, and converting the data in the data set into an original feature vector; the original feature vectors are input into a plurality of Transform models with different hyper-parameter settings for model training, a linear transformation module is included in front of each Transform model, and a plurality of cardinal number prediction results are obtained through training; and selecting a model with a minimum error as an optimal model according to a cardinality prediction result to obtain all original query feature vectors and corresponding optimal models, and learning a mapping relationship between the query features and the optimal models by utilizing a classifier. Obtaining a new SQL query input classifier, and distributing an optimal model based on the index data set; and processing the feature vector by using the optimal model to obtain a cardinality estimation result. The defect of single model query is changed, and the accuracy of a query result is improved based on feature classification query.

The invention discloses a high-concurrency and sequential data processing method for batch import class interfaces. In a request verification and forwarding stage, request data are verified, a guarantee sequence is locked based on a merchant ID after verification is passed, then the data are batched, request and batch information is persisted to a database, batch information is delivered to a corresponding message queue Topic according to a service scene, and the request ID is returned. In the consumption queue data stage, a consumer thread subscribes to messages of the message queue, processes the messages, updates a batch state, manually submits consumption offset, and dynamically adjusts the sleep duration of the thread according to a system load to realize intelligent peak clipping. And a progress result query stage: extracting data from the database based on the request ID, calculating a processing progress and a failure reason, formatting and returning the processing progress and the failure reason, and supporting front-end visual display. Through the message queue peak clipping, the differentiation sequence control strategy and the multi-dimensional data persistence, the high concurrency bearing capacity of the interface is improved, and the data consistency and reliability are guaranteed.

A computer system of a vehicle configured to monitor vehicle surroundings is provided. The computer system comprises processing circuitry configured to acquire data samples from a monitoring device configured to measure a distance to an object located within a field of view of the monitoring device, determine a difference between at least two sequential data samples of the measured distance from the monitoring device to the object, and if said difference exceeds a set first threshold value, determine a number of distances within an operating range of the monitoring device to the object that are not measured by the monitoring device during a set time period. The measured distance to the object is not relied upon if the number of distances that are not measured during the set time period is below a set second threshold value.

It is desired to provide a technique for accurately estimating model parameters by data assimilation.SOLUTION: And a data assimilation unit configured to obtain a data assimilation result by executing sequential data assimilation on the basis of observation data obtained by observing a state of an observation target, the simulation result, and a variance of system noise in the simulation unit, in which the data assimilation unit makes a variance of system noise at an end of the sequential data assimilation smaller than a variance of system noise at a start of the sequential data assimilation.SELECTED DRAWING: Figure 1

This invention provides a lossless compression and decoding method for depth images. Lossless compression includes: dividing sequential data into multiple independent image blocks according to predetermined partitioning rules; selecting a corresponding thread model based on the characteristics of the depth image; calling a thread based on the thread model and using a reversible encoding algorithm to independently encode each image block; wherein, at the start of encoding for each image block, the encoder state is reset so that the encoding of each image block does not depend on the data of the preceding image block; generating metadata containing position index and compressed data length information and assembling it with the encoded image block for output. Lossless decoding includes: reading the metadata and decoding at least one image block based on the metadata. Through the above methods, parallel encoding and decoding are achieved; precise positioning via physical offsets and random access and on-demand region decoding are supported, improving the processing response of high frame rate depth streams under low computing power environments while ensuring lossless restoration.

System receives data stream, groups sequential data segments associated with data stream until initial sequence of data segments is formed which is larger than minimum size. System groups sequential data segments which are next after initial sequence of data segments until next sequence of data segments is formed which combined with initial sequence of data segments is larger than maximum size. System determines feature value for each data segment in next sequence of data segments. System selects value from feature values, and data segment corresponding to selected value. System divides next sequence of data segments at selected data segment into part of initial group of data segments and part of next group of data segments. System combines part of initial group of data segments with initial sequence of data segments as initial group of data segments. System routes initial group of data segments or group of corresponding fingerprints for deduplication.

A data storage device includes a memory device and a controller coupled to the memory device. The controller is configured to interact with a host device using a Universal Flash Storage (UFS) interface protocol, provide a hint to the host device, switch between a first mode and a second mode, retrieve data from the memory device, and transfer the data to the host device. The hint includes an indication of what order data is to be received from the data storage device. After the hint is provided, the order of the data will be a different order than a requested order.

Methods, systems, and devices for physical prefetch are described. A memory system may prefetch data from physical addresses that are sequential to a physical address that has recently been read by the memory system. For example, the memory system may determine that a rate of access for a virtual block exceeds a threshold, and the memory system may prefetch the physically sequential data in response to the rate of access exceeding the threshold. In some examples, the memory system may receive a command to read a logical block address corresponding to a first physical address. The memory system may determine that the rate of access to the virtual block exceeds the threshold. The memory system may identify one or more second physical addresses that are sequential to the first physical address, and the memory system may transfer data from the one or more second physical addresses to a buffer.

This invention discloses a method and system for processing terminal security information in wireless networks, relating to the field of terminal security information technology. The method includes collecting a physical layer feature set and an internal state feature set of the terminal; preprocessing the features using a lightweight edge computing module; arranging the physical layer feature set in chronological order to construct sequential data; recursively calculating the mean and variance of the features to generate a real-time feature baseline range; combining the terminal's internal state feature set with a weighted correction of the real-time feature baseline range to obtain a dynamic baseline range; triggering an alarm based on the dynamic baseline range; triggering re-authentication; and after three consecutive alarm triggers, feeding back to the LSTM model for online updating. This method overcomes the limitations of traditional single-feature authentication, avoids the misjudgment of legitimate signals as abnormal due to multipath effects, reduces the false positive rate through dynamic threshold adjustment, and reduces redundant operations by 30%.

Systems and methods create decision trees from transformer weights for the non-tabular and sequential data. For example, the system may take initial non-tabular and / or sequential data convert it into a binary variable (e.g., is it there or not). The system may then extract features out of the model that is learned on the data as opposed to extracting features itself. The system may then fit a decision tree on these extracted features to generate an approximation of the transformer that is interpretable / explainable.

The present invention concerns a method for detecting faults in mechanical components by classifying text strings describing issues with said mechanical components by combining deterministic, rule-based keyword extraction algorithm with deep learning models. It uses a Generative AI model to generate synthetic data, especially for underrepresented classes. The system consists of two main components: a probability map that implements a deterministic, rule-based keyword extraction algorithm by focusing on negations and relevant keywords, and a Transformer-based model that handles long-range dependencies in sequential data. The outputs from both models are combined using score-based aggregation, and an ensemble classifier is trained to integrate these outputs, leading to improved accuracy in handling diverse data. The invention also includes a system with modules for data collection, preprocessing, augmentation, training, and evaluation to ensure optimal classification performance.

The invention discloses a golf glove magic tape accessory with distance measurement positioning and ball hitting data recording, which comprises an operation key module used for generating a trigger signal; the positioning unit is used for generating a distance data signal; the sensor module is used for generating a standard sensing signal; the data transmission unit is used for receiving the court map data and the hole sequence data and sending the ball hitting record data; the memory is used for storing standard sensing signals, court map data and hole sequence data; the display module is used for displaying a distance value in the distance data signal; the controller is used for sending a control signal; the clock unit is used for providing a clock signal; and the power management unit is used for providing power. Therefore, the system has the advantages that a distance measurement positioning function and a ball hitting data acquisition function are integrated, customized court map and hole sequence data can be received, ball hitting record data can be uploaded, and distance measurement accuracy and data acquisition pertinence are improved.

The invention discloses an SSD garbage collection acceleration method and device based on data distribution, equipment and a medium, and relates to the technical field of garbage collection, and the method comprises the following steps: distinguishing sequential data and random data when a host writes in, and maintaining a random data logic address table for a physical block to record a written random data logic address; when the physical block is fully written, if the effective data count is not greater than the number of the logic addresses recorded in the table, checking the current mapping of the addresses in the table; if the number of the addresses still pointing to the block is equal to the effective data count, only data corresponding to the addresses are moved, and then the physical block is recycled. According to the method, through logical address record and mapping comparison, a small number of residual random data can be accurately positioned and moved in a specific scene, the overhead of traversing a whole physical block for data recovery in a traditional scheme is avoided, meanwhile, a single write-in point is kept, and the data recovery efficiency is improved. And the problems of complex logic, difficult recovery, increased space occupation and the like caused by maintenance of multiple write-in points in a traditional shunting scheme are solved.

Methods and systems for identifying malicious tenants using machine learning (ML) models and spatialized activity data within a cloud platform. A method includes accessing sequential activity data for a tenant, converting the sequential data into spatial data by creating multi-dimensional representation(s) expressing the sequential data across different timescales, training an ML model based on the spatial data, and deploying the trained ML model in the cloud platform to detect tenants engaged in malicious activities. A method also includes accessing sequential activity data for a tenant, converting the sequential data into spatial data by creating multi-dimensional representation(s) expressing the sequential data across different timescales, inputting the spatial data into a trained ML model, and initiating a remedial action against the tenant based on an output indicating that activity data for the tenant has a threshold probability of being malicious.

Methods, systems, and devices for logical address indications for accessing sequential data from a memory system are described. The described techniques provide for a memory system to access sequential data associated with non-contiguous logical address ranges based on receiving a read command including information indicating the non-contiguous logical address ranges. For example, the read command may indicate the non-contiguous logical address ranges in an extra header segment (EHS) data field and the memory system may use the indication of the non-contiguous logical address ranges to improve access performance, such as by informing a pre-reading algorithm. In some examples, the memory system may identify that the EHS data field includes a logical address table based on one or more second fields of the read command, such as an EHS length field, an EHS type field, and an EHS sub-type field.

A low power Bidirectional Encoder Representations from Transformers (BERT) model natural language processor is provided. The methodology includes: creating a set of non-sequential data points from user data including numerical features; first binning overlapping types of numerical features from the non-sequential data points into consecutive numbers; first allocating the consecutive numbers as binned into a plurality of quantiles; first assigning a first category token to each of the plurality of quantiles, each of the first category tokens being BERT compatible; creating a time ordered sequence of transactions of the user including text features and numerical features; second assigning second category tokens to the time ordered sequence of user transactions; concatenating the first category tokens and the second category tokens into a sequence; pre-training the BERT model based on the sequence; and submitting a query including text features and numerical features to the trained BERT model.

The application provides an integrated learning classification method and device based on an online sequential pre-interference layer, and relates to the technical field of data stream classification.The application obtains historical vibration sequential data and real-time online vibration sequential data; designs an integrated classifier composed of multiple heterogeneous base classifiers; the base classifier is a four-layer feedforward neural network model, an interference layer is added between the input layer and the hidden layer of the OS-ELM network to realize nonlinear kernel mapping of online sequential samples; each base classifier of the integrated classifier is initialized and learned sequentially based on the obtained historical vibration sequential data and online vibration sequential data; and the newly obtained online vibration sequential data is classified and predicted based on the integrated classifier.The application can effectively solve the concept drift and class imbalance problems in dynamic data streams, and improve the recognition ability of the model for minority class samples and the adaptability of the model to data distribution changes.

The embodiment of the application provides a method and device for automatically generating an Internet of Things project end to end based on a large language model. The method comprises: obtaining a project requirement text; calling a large language model to analyze the project requirement text to obtain structured requirement information of the Internet of Things project; in a preset skill warehouse, finding a skill module that is semantically matched with the structured requirement information as a target skill module; analyzing the compatibility and dependency relationship between each target skill module to obtain the calling sequence of each target skill module and the data flow direction between each target skill module; calling the large language model to generate a target project code based on the structured requirement information, the calling sequence, the data flow direction, and the description information of each target skill module; and in the case that a confirmation instruction for the target project code is received, generating a version baseline of the Internet of Things project based on the target project code. The project development efficiency can be improved.

Disclosed in the present invention is a learning management system for predicting concentration on the basis of data generated from a metaverse teaching environment. The present invention may provide a learning management system based on data generated from a metaverse teaching environment, which includes a sequence model for predicting a learner's concentration from sequential data regarding the learner's viewing posture. The learning management system is capable of maintaining the learner's concentration by switching, replacing, or reproducing video content for which a relatively low concentration is predicted with video content that is capable of arousing the learner's interest. For example, by transmitting image information or numerical information related to the learner's viewing posture to a terminal of a controller, appropriate learning guidance, such as switching video content for which a relatively low concentration is predicted, may be provided, thereby maintaining the learner's concentration and improving learning achievement.

The invention relates to a method and a computing system for improved temporal processing of sequential data. Systems and methods may include or utilize a machine learning model (e.g., a convolutional neural network) that includes one or more temporal residual connections. In particular, each temporal residual connection may respectively supply one or more intermediate feature data sets generated from a current sequential input by a current instantiation of the model to one or more other instantiations of a machine learning model applied to process one or more other sequential inputs. For example, other instantiations of the machine learning model may include a subsequent instantiation of the machine learning model applied to one or more subsequent sequential inputs in the processing sequence that follow the current sequential input, and / or applied to previous instantiation of the machine learning model in the processing sequence that precedes one or more previous sequential inputs in the current sequential input.

The invention discloses a liquid cooling host fault early warning method and device based on cloud computing. The method comprises the following steps: acquiring multi-source sensor data at a liquid cooling host end and uploading the data to a cloud platform, and performing cross-source time alignment and compensation processing to obtain preprocessed time sequence data; performing anomaly detection and trend analysis based on the sequential data to identify an abnormal pattern; inputting the abnormal mode into a machine learning model to obtain a fault prediction result and diagnosis information; according to the prediction result and the diagnosis information, a reinforcement learning algorithm is adopted to generate a parameter optimization scheme; and finally, based on the optimization scheme, automatically triggering early warning and executing a control strategy. According to the invention, accurate prediction and early warning of the fault of the liquid-cooled host are realized, the fault identification accuracy is improved through multi-source data fusion analysis and a machine learning model, intelligent optimization of parameters of the liquid-cooled system is realized by adopting a reinforcement learning algorithm, automatic execution of an early warning control strategy is realized through a cloud collaborative architecture, and the accuracy of fault identification is improved. And the reliability and the operation and maintenance efficiency of the liquid cooling system are remarkably improved.

Described herein is a system for syncing asynchronously received sequential data from disparate sources. In an embodiment, a central system may receive data transmissions from disparate sources. Each data transmission includes a timestamp and an identifier of the disparate source. The central system may sort the data from the data transmissions of disparate sources in chronological order based on the timestamps. The central system may group the data based on the identifier of the disparate source and normalize the data of the data transmissions to be in a specified format. The central system may store the normalized data of the data transmission in a data storage facility based on the identifier of the disparate source and the sorted order of the data.

This application relates to methods, systems, and non-transitory computer-readable media associated with conditional object-centric slot attention learning for video and other sequential data. One method includes obtaining a first feature vector and a second feature vector respectively representing content of first and second image frames of an input video. The method can also include generating, based on the first feature vector, first slot vectors, where each slot vector represents an attribute of a corresponding entity as represented in the first image frame, and generating, based on the first slot vectors, predicted slot vectors including corresponding predicted slot vectors representing a transition of the attribute of the corresponding entity from the first image frame to the second image frame. The method can also include generating, based on the predicted slot vectors and the second feature vector, second slot vectors including corresponding slot vectors representing the attribute of the corresponding entity as represented in the second image frame, and determining an output based on the predicted slot vectors or the second slot vectors.

Disclosed embodiments may provide techniques for performing database operations using a hierarchy-enhanced B-tree index. A computer-implemented method can include executing a query to access a data item by traversing a root node of a hierarchy-enhanced B-tree index to a particular leaf node connected to the root node. In some instances, the particular leaf node indicates a first unstructured-data record. The method can also include determining that the particular leaf node includes a sequential data structure. The sequential data structure identifies a hierarchical relationship between the first unstructured-data record and other unstructured-data records indicated by one or more other leaf nodes. The method can also include using the sequential data structure to access the one or more other leaf nodes without additional path traversals of the hierarchy-enhanced B-tree index. The method can also include retrieving the data item by parsing the other unstructured-data records.

The disclosure configures a memory sub-system controller to store random data in a different layout from sequential data. The controller receives a request to store a set of data to a set of memory components. The controller determines whether the set of data corresponds to either sequential data or random data and selects a write cursor from a plurality of write cursors to associate with the set of data in response to determining whether the set of data corresponds to the sequential data or the random data. The controller programs the set of data to one or more of the set of memory components according to a data layout associated with the selected write cursor.