202 results about "Vector generation" patented technology

Methods for video decoding and encoding, apparatuses, and non-transitory computer-readable storage media are provided. In one method, a decoder may obtain a first block vector (BV) and a second BV based on a bi-predicted intra block copy (IBC) mode. Additionally, the decoder may generate a first prediction block based on the first BV and generating, by the decoder, a second prediction block based on the second BV. Furthermore, the decoder may obtain a final prediction block for a current block based on the first prediction block and the second prediction block.

This application provides a method, apparatus, electronic device, and storage medium for determining responses to questions, belonging to the technical field of question response determination, and applicable to scenarios in financial, medical, health, and elderly care insurance application systems. The method includes: responding to a question input by a user, performing semantic and compliance constraints on the question, and generating a structured control vector corresponding to the question; inputting the structured control vector into multiple perturbation decoding heads, and generating candidate response sequences corresponding to the business semantic distribution of the question and each perturbation decoding head based on the structured control vector; wherein the business semantic distributions of the multiple perturbation decoding heads are different; and determining the target response sequence for the question based on the candidate response sequences. This application improves the diversity and accuracy of responses by constructing a structured control vector through pre-generation semantic and compliance constraints, and by generating candidate responses in parallel using perturbation decoding heads with differentiated business semantic distributions.

This invention discloses a spline-adaptive method for eliminating power frequency interference in surface electromyography (EMG) signals, within the field of EMG signal noise reduction technology. The method includes the following steps: frequency domain input vector generation and frequency domain spline adaptive filtering; power frequency interference elimination; construction of a semi-quadratic criterion cost function and filter coefficient update; iterative update, repeating the above steps until noise reduction is complete. This spline-adaptive method for eliminating power frequency interference in EMG signals employs a robust frequency domain implementation method to construct the spline adaptive noise reduction system, exhibiting good stability, high computational efficiency, and the ability to efficiently eliminate power frequency interference.

This invention discloses an emotion-driven multi-region adaptive joint optimization intelligent temperature control method for open spaces. The method includes establishing a correspondence between regions and temperature control execution units, acquiring region state data, calculating the region's emotional stress index and its corresponding confidence level, establishing an inter-regional influence weight matrix to obtain coupled emotional stress indices for each region and determining influence coefficients, constructing a multi-regional joint optimization model, determining whether emotional event triggering conditions are met, solving the multi-regional joint optimization model to obtain the optimal control vector, generating and issuing control commands, and acquiring environmental parameters and temperature control interaction disturbance parameters after execution in the next control cycle, updating the method based on evaluation results. The beneficial effects of this invention are: it can achieve timely, coordinated, and stable temperature control in open, multi-user shared spaces, improving comfort and reducing fluctuations caused by human intervention.

The application provides a kind of adaptive human-computer interaction method and system based on multi-modal perception, it is related to human-computer interaction technical field, acquires the multi-modal original perception information flow that user obtains in mixed reality space through depth vision, bone conduction and flexible tactile sensor, respectively to its penetration depth calculation, voiceless speech decoding and mechanical response matching processing are handled, obtain interactive penetration depth sequence, internal speech instruction coding sequence and virtual tactile feedback expected intensity sequence.Multiple modal intention conflict detector detects cross-modal intention conflict, generates intention conflict representation vector, finally generates interactive mode smooth switching instruction according to the intention conflict representation vector, triggers mixed reality space interaction paradigm conversion, realizes the adaptive conversion of interaction paradigm in mixed reality space, effectively solves the problem of multi-modal interaction intention conflict processing difficulty in traditional method, greatly improves the accuracy, fluency and naturalness of human-computer interaction.

This application belongs to the interdisciplinary field of new energy power generation and meteorological forecasting, specifically relating to an intelligent prediction method and system for photovoltaic snow accumulation loss. The method includes: real-time acquisition and fusion of multi-source data from satellites, ground stations, and the power plant itself to form a spatiotemporally aligned feature dataset; calculation of key physical features such as the equivalent optical thickness of the snow layer and power attenuation rate based on this dataset, and inputting these features into a pre-trained identification module to dynamically determine the current snow accumulation stage; inputting time-series data containing physical features into the encoder of a pre-trained physics-AI hybrid prediction model to extract a high-dimensional latent vector representing the current comprehensive state; and intelligently selecting the corresponding dedicated prediction sub-module based on the identified snow accumulation stage label to generate an accurate prediction sequence of photovoltaic power output loss rate for future periods based on the latent vector. This method achieves multi-source information fusion, physical mechanism embedding, and process adaptive prediction.

The application relates to the technical field of artificial intelligence-based music melody automatic generation systems, and particularly discloses an artificial intelligence-based music melody automatic generation system. The system comprises an intention analysis and feature mapping module, which is used for converting a user intention into a structured semantic feature; an emotion evolution state machine module, which is used for dynamically generating a time-series emotion state vector according to the semantic feature; a hierarchical controllable generation network module, which is used for generating a note sequence with specific styles, coherent emotions and diversity based on the semantic feature and the emotion state vector; and a post-processing and music theory constraint module, which is used for performing compliance correction and optimization on the note sequence. The application can generate a melody with coherent emotion evolution logic, high matching of a user intention, various styles and compliance with music theory.

This invention discloses a vehicle data classification and grading system based on multidimensional compliance vector analysis, comprising: a data receiving and preprocessing module that receives and stores vehicle data list files to obtain a flattened data record table; a compliance vector generation module that performs semantic parsing to obtain a compliance vector sequence; an improved ETSformer analysis module that constructs an improved ETSformer model and generates analysis results; a compliance distribution measurement module that uses Wasserstein distance to obtain distance measurement results; a hierarchical priority arbitration module that determines labels and records the rule basis and model results; and a result reconstruction module that aggregates data according to context identifiers to obtain data classification and grading results. This invention achieves automatic classification, grading, and distribution consistency calibration of complex vehicle data lists by improving the ETSformer model and Wasserstein distance measurement structure.

The application relates to the technical field of radar detection, in particular to a pulse compression ultra-low sidelobe nonlinear frequency modulation waveform design method based on dimension layering and two-stage optimization, which comprises the following steps: given signal parameters, GA parameters and a coefficient vector; the coefficient vector is initialized, a global search is performed by using a GA algorithm, genetic operations are performed, and after being crossed and mutated, energy projection normalization is performed to meet an energy constant constraint; when a termination condition is met, the global search is stopped and a candidate solution set is output; on the basis of the candidate solution set, a parameter space distance threshold is introduced to perform diversity screening, K candidate solutions which are dispersed from each other and excellent in performance are extracted and are zero-padded and dimensioned to obtain K seed vectors; the seed vectors are taken as initial values, a quasi-Newton method is respectively run to perform local optimization and finishing, a coefficient vector corresponding to an initial value which makes the sidelobe energy minimum is selected as an optimal coefficient vector, and a nonlinear frequency modulation waveform with the characteristics of an ultra-low sidelobe is generated based on the optimal coefficient vector.

The application discloses a Kalman filtering state decoupling and parallel computing method for multi-target tracking, and comprises the following steps: S1, generating several two-dimensional Kalman subsystems according to the original high-dimensional state space and column vectors of Kalman filtering hardware; S2, constructing two-dimensional Kalman filtering processing units for each two-dimensional Kalman subsystem, and outputting a high-dimensional state vector; and S3, generating a high-dimensional target state estimation result according to the high-dimensional state vector. When the multi-target tracking algorithm is applied, the parallel two-dimensional Kalman filtering processing array is simultaneously executed in the FPGA, and the processing delay of single filtering is greatly shortened. The soft and hard collaborative architecture completely sinks the heavy matrix operation to the FPGA, significantly reduces the CPU load, enables the CPU to focus on upper logic control, and thus realizes real-time multi-target tracking with high frame rate and low delay on the end-side heterogeneous platform.

The application discloses a kind of green ecological monitoring management systems based on knowledge graph, including multimodal data reading module: generate first preprocessed data and second preprocessed data;Multimodal feature extraction module: first preprocessed data is input to ShuffleNet model, and second preprocessed data is input to RoBERTa model;Ecological space topology construction module: constructs ecological space topology atlas;Space vector generation module: based on two-stage space mapping mechanism, generates visual space vector and text space vector;Space probability model construction module: constructs space probability model;Cross-modal space inference and matching module: execute cross-modal causal consistency inference;Ecological knowledge graph generation module: realize the knowledge graph storage and update of multimodal entity pair.The application realizes higher precision multimodal fusion and more complete knowledge graph construction effect in complex ecological monitoring scene.

This application relates to a content search method, apparatus, electronic device, and storage medium. The method includes: responding to a search request for a target object, using a generative large model to perform text correction processing on initial search terms carried in the search request to obtain corrected search terms; using the generative large model to search for target media content matching the corrected search terms from a preset vectorized knowledge base as first search context information, the preset vectorized knowledge base being used to store vectors of content vertical domain knowledge; guided by the first search context information, the generative large model performing search term update and rewriting processing on the corrected search terms to obtain corresponding rewritten search terms; and performing content search based on the rewritten search terms to obtain content search results. According to the technical solution provided in this application, the generative large model is made more lightweight, and the rewritten reference is more comprehensive, thereby making the search rewriting more accurate.

This invention belongs to the technical field of server resource scheduling and management, and relates to a cascaded scheduling system for server AI computing power modules for elastic computing. This system utilizes a situational awareness unit deployment module to collect micro-level operational data from the modules and generate raw sampling information; a real-time situational vector generation module constructs a multi-dimensional real-time situational vector; a global pressure map construction module maps the vector to generate a global cascaded pressure map for the pipeline; a dynamic bottleneck diagnosis module identifies abnormal nodes and determines hard bottlenecks and pseudo-blockages; a collaborative flow control strategy orchestration module generates push-pull linkage control commands; a predictive performance state adjustment module generates power level adjustment commands based on prediction time; and a command issuance and feedback closed-loop module drives execution and corrects parameters. This invention solves the problems of difficult accurate diagnosis of micro-level performance bottlenecks and the lack of global collaborative dynamic adjustment capabilities for data flow and power consumption states in heterogeneous AI computing power module cascaded systems.

The application discloses an unmanned distribution task allocation method based on multi-agent cooperation, and specifically comprises the following steps: step 1, obtaining distribution task data and unmanned distribution body running data; step 2, constructing a multi-agent state vector set; step 3, generating a task cost disturbance variable set and an execution time delay disturbance variable set; step 4, establishing a disturbance variable parameter coupling relationship, constructing a non-tensor polynomial chaos expansion model, and generating a disturbance propagation risk vector; step 5, constructing a parameter coupling manifold based on the expansion coefficient, and screening candidate parameters by using a DIRECT algorithm; and step 6, generating a task allocation score matrix based on the candidate parameters and the disturbance propagation risk vector, and outputting a multi-agent task matching result. The application introduces parameter coupling modeling to improve the stability of cooperative allocation.

This invention provides a traffic sign recognition method and related apparatus. The method includes: acquiring real-time image data and environmental data collected during vehicle movement; concatenating the preprocessed environmental data into an environmental vector, and generating a modulation vector based on the environmental vector; applying channel attention modulation to a feature map using the modulation vector to obtain a modulated multi-scale image feature map; performing feature recalibration processing on the multi-scale image feature map to obtain a recalibrated feature map; performing traffic sign detection based on the recalibrated feature map and outputting an interpretable detection result for the traffic sign; and performing online incremental learning on the target detection model based on the interpretable detection result according to preset triggering conditions, and collaborating with a cloud server through federated aggregation. Based on this, this invention can intelligently recognize traffic signs, obtain interpretable detection results, and support online incremental learning.

The application relates to a multi-source data fusion operation control method and system, and relates to the technical field of port automation and intelligence. The method comprises the following steps: acquiring multi-source operation state data of ship unloader grab operation, extracting a feature vector, and generating an operation state matrix; based on a dynamics model of the ship unloader grab and the operation state matrix, a standard state equation of the ship unloader grab containing state variables and control variables is constructed; under the dynamic relationship defined by the standard state equation, a multi-objective function containing a path optimization target and an energy consumption control target is constructed, and a solving constraint condition of the multi-objective function is set; a non-dominated sorting genetic algorithm is used to optimize and solve the multi-objective function under the solving constraint condition, and an optimized control result of the grab operation is generated; and the optimized control result is issued to an executing mechanism of the ship unloader to control at least one ship unloader to execute the grab operation. By using the application, the automation level and operation efficiency of the ship unloader grab operation can be improved.

The application relates to the technical field of robot control, and discloses a terminal position dynamic compensation method based on fusion of visual servoing and force feedback, which comprises the following steps: collecting a top view image, segmenting an irregular residual hole contour, and calculating a dynamic alignment center coordinate; then, a reverse conical spiral descending trajectory is generated in combination with a lateral force vector to guide a fiber tail fiber connector into a guide port; in the insertion stage, the curvature radius characteristic of a force-displacement curve resistance peak point is used to determine the authenticity of bottom touching; for the case that the bottom touching is determined to be false, the side projection chord length is used to deduce an under-insertion depth, and a high-frequency micro-oscillation compensation insertion is driven until a real buckle locking feature is captured; the application realizes accurate alignment and reliable positioning of fiber plug-in and plug-out actions under complex working conditions such as shielding and mechanical wear.

The application discloses an exoskeleton robot control parameter determination method and system, and particularly relates to the technical field of robot control, which comprises the following steps: synchronously collecting rigid joint torque, flexible cable tension and inertia signals, generating a double-channel response spectrum through Fourier transform and obtaining a frequency ratio vector; calculating kinetic energy coupling indexes and viscous lag indexes, generating a bandwidth difference index through a collaborative phase model combined with an inertia-damping signature vector; adaptively obtaining a rigid gain matrix and a flexible gain matrix under energy consumption constraints by using sparse Bayesian optimization; constructing a weight function according to action prediction phase and energy consumption constraints, synthesizing a composite control instruction stream for the two gain matrices; monitoring expected and measured torque residual errors, and when the errors exceed a threshold, generating a gain increment according to gradient sensing and rewriting a coupling feature extraction unit to realize millisecond-level gain correction, thereby solving the problem that the prior art cannot accurately quantify energy mutual feedback and phase lag, leading to coupling oscillation caused by unified gain mismatch.

This invention discloses an AI-based integrated system for intelligent classification, storage, and retrieval of archives. The system includes: an archive classification module that inputs archive text into a MacBERT model for classification; a category template generation module that determines the target entity field identifier set based on archive category labels; a named entity recognition module that generates structured archive metadata; a multi-view encoding module that generates view pooling vectors; a multi-view interactive sentence vector generation module that generates sentence vector representations of target archives; a vector index library module that stores sentence vector representations of target archives; a structured database module that stores structured archive metadata; and an archive retrieval module that outputs retrieval results by combining structured archive metadata during the retrieval phase. This invention is applicable to automated management and semantic-level intelligent retrieval scenarios for large-scale government and enterprise archives.

This invention discloses a method, apparatus, medium, and product for path and trajectory co-optimization based on neural network enhancement. The method includes obtaining the start and end nodes of a moving target and the graph structure of the problem to be solved; obtaining a robust Pareto front of the problem to be solved based on a co-optimization model; training a variational autoencoder network model to generate a candidate path set; obtaining ideal velocity curves based on the candidate path set; generating a trajectory set using a trajectory generation model; mutating the latent space feature vectors of the variational autoencoder network model to obtain new latent space feature vectors; generating a new candidate path set based on the new latent space feature vectors; obtaining the ideal velocity curves of each candidate path in the new candidate path set to generate a new trajectory set; and filtering the trajectory set and the new trajectory set to obtain a robust effective trajectory set as the robust Pareto front. This invention effectively solves the shortcomings of current multi-objective robust co-optimization methods for paths and trajectories.

The application relates to the technical field of Internet of Things device power consumption management, and discloses a self-adaptive dynamic power consumption adjustment method and system for an Internet of Things device, which comprises the following steps: acquiring real-time environment parameters and operation parameters of the Internet of Things device, and generating a current multi-dimensional state vector through coding; inputting a multi-dimensional state space, matching and determining a current state node, and extracting a set of bound basic hardware control parameters thereof; acquiring adjacent nodes of the current node, extracting a historical state transition path thereof, and calculating a parameter offset trend vector; fusing the set of basic parameters and the trend vector, generating a target instruction set, and issuing an adjustment; collecting actual energy consumption and environment parameters, calculating an execution deviation value, correcting current and adjacent node coordinates according to the deviation value, and updating the set of basic parameters. The application improves the power consumption efficiency, environmental adaptability and long-term operation stability of the Internet of Things device under complex and changeable environments.

This application discloses an AI-based cross-scene furniture e-commerce interaction system and method. The system includes: an event receiving and recording module, a sequence verification module, a representation vector generation module, a credibility calculation module, an attribution classification module, and a sample update module. The system receives interaction events, establishes room scene records and product version records, and generates version influence vectors; it performs sequence verification on the same user events, constructs a time series diagram, and generates representation vectors; it calculates event credibility by combining scene validity period, product version validity period, time deviation, and associated anomalies; it performs replay verification and attribution classification on low-credibility negative feedback events, and updates training samples, user preference states, and product recommendation states accordingly. This system can improve the accuracy of interaction data processing and recommendation updates.

The application provides a long-short-term combined reinforcement learning charging load prediction method and system, and belongs to the technical field of intelligent power distribution network optimal operation. The method comprises the following steps: acquiring multi-dimensional charging feature data of multiple charging stations in a region at continuous time steps to form a three-dimensional data matrix; extracting a spatial feature vector of each charging station; extracting a time feature vector of each charging station; generating a space-time fusion feature vector; constructing a multi-agent reinforcement learning prediction model, each charging station is defined as an agent, a global state vector is generated based on the local observation state of each agent and the global power grid state information, and the local value function of each agent is combined through a hybrid network to generate a global value function under the condition of meeting the monotonicity constraint to guide each agent to output a charging load prediction value at the next moment. The application deeply fuses the space-time features of the charging load, realizes multi-site collaborative prediction, and improves the prediction accuracy and robustness.

This invention belongs to the technical field of process control and automation, and relates to an intelligent switching system for multi-stage reaction process parameters in a reactor. The system includes: a transient physical sequence generation module to acquire signal-generated variable sequences; a Jacobian tensor array construction module to construct a dynamic heat flux Jacobian tensor array reflecting temporal coupling strength; a system topological singularity identification module to analyze the array and generate eigenvalue reduction mutation flags; a hardware wake-up vector generation module to generate a hardware-level execution clock trigger vector for the wake-up controller in response to the flags; a transient residual data extraction module to extract transient heat flux residual data fragments via a direct memory access mechanism; and a control state overwrite activation module to overwrite the controller state using the fragments and output a disturbance-free control command for the target stage. This invention solves the problem of difficulty in accurately capturing physical phase transition singularities and the initial state of the synchronous controller during multi-stage operation, leading to control disturbances during switching.

The application discloses a commodity semantic representation vector generation method and device, and relates to the technical field of electronic commerce. A specific embodiment of the method comprises the following steps: performing word piece covering processing on each commodity title in a commodity title set respectively, obtaining a word piece index vector, a sample length vector and a text segment vector corresponding to each commodity title according to the commodity title after the word piece covering processing, and inputting the vectors into a pre-training model to obtain a first vector representation corresponding to each commodity title and a covering prediction result; then, adjusting parameters of the pre-training model to generate a semantic representation vector extraction model; and extracting vectors from the commodity title set by using the semantic representation vector extraction model to obtain a commodity semantic representation vector. The embodiment is more suitable for commodity language representation vector extraction, and can obtain a more accurate semantic representation vector, improves the semantic representation effect, and improves the use effect of the commodity semantic representation vector.

Disclosed are an apparatus and method for providing a speech video. The apparatus for generating a speech video according to an embodiment comprises: an encoder for encoding a person image to generate an identity latent vector; a voice emotion recognition model for recognizing emotion from speech audio; a motion latent vector generation model which is a probabilistic generative model for generating a motion latent vector on the basis of the speech audio and the emotion; and a decoder for generating a speech video of speech spoken by the corresponding person on the basis of the identity latent vector and the motion latent vector.

This application relates to a knowledge graph-based method, system, and electronic device for engineering data asset management, belonging to the technical field of electronic data processing. The method includes the following steps: acquiring and packaging raw industrial data; inputting it into a unified representation platform for multi-source heterogeneous industrial data; generating standardized multi-source homogeneous industrial data; constructing an initial knowledge graph and decoupling it into first and second local information graphs; calculating the global embedding representation result; constructing a hierarchical multi-team intelligent module; decomposing the processing target into multiple sub-tasks through a decision-level intelligent agent and assigning them to professional-level intelligent agents; and iteratively optimizing the knowledge graph. This application achieves the goals of data form standardization and deep semantic feature extraction, realizing the technical effects of multi-level information fusion and global representation vector generation. It solves the problems of information silos and insufficient inference accuracy caused by complex industrial field data acquisition protocols and insufficient knowledge association.

The application relates to the technical field of large models, and provides a prompt vector generation method, device, equipment and medium, which extracts and embeds the multi-modal information of a target user, generates a personalized query vector carrying the personalized features of the target user, and fuses relevant information of the personalized query vector and a pre-constructed industry knowledge prompt library to generate a target prompt vector of the target user. The target prompt vector is determined in combination with the personalized features of the target user and global user common features, and the personalized and customized processing of the prompt vector is realized. The application can be applied to a financial scene, and when a downstream task such as insurance recommendation generates a personalized execution result for a user according to the target prompt vector, the personalized features of the target user and the global user common features can be simultaneously referred to, the target user's personalization is considered, and the rules or trends gradually presented in the insurance business scene are not deviated from, so that the execution effect of the downstream task is improved.