78 results about "Score vector" patented technology

The invention discloses an intelligent education robot question answering system based on voice recognition and a knowledge graph, and particularly relates to the technical field of artificial intelligence. The method comprises the following steps: performing feature extraction on a user voice signal to obtain a text sequence and a multi-modal context feature; recognizing subject domain judgment and question answering intentions based on supervised classification and keyword rule fusion, and outputting a subject domain prior probability and a question answering target vector; determining polysemy words in the text sequence by using a context window, generating a semantic item sequence and semantic item confidence, constructing a subject domain sub-graph based on a subject domain prior probability, and obtaining a candidate reasoning path and a path scoring vector; determining a target teaching concept and an optimal reasoning path by combining Bayesian inference and consistency verification; generating a personalized question answering result aiming at the question asked by the user through fact retrieval and knowledge derivation in combination with the question answering target vector; accurate and efficient intelligent teaching question answering can be realized, and question answering accuracy and intelligent interaction capability of the education robot are effectively improved.

The invention relates to the technical field of airport operation management, in particular to a data-driven airport scene sliding path intelligent generation method, which comprises the following steps: acquiring a plurality of taxiway sequences to form a taxiway sequence set; establishing a neural network model and a target function, and training the neural network model to obtain a taxiway embedded vector generator; starting and ending point pair coding, environment feature coding and related flight coding are carried out based on the taxiway embedded vector generator, and heterogeneous fusion features are obtained; carrying out autoregression generation based on Transform to obtain a plurality of candidate taxiways and corresponding score vectors; the multiple candidate taxiways are expanded and screened according to the score vectors of all the candidate taxiways, multiple candidate paths are generated, and the high-quality airport scene sliding path can be generated.

The invention relates to the technical field of natural language processing, in particular to a retrieval system and method based on retrieval enhancement generation, and the method comprises the steps: obtaining a user query text through a user interaction interface, and carrying out the embedding processing of the user query text to obtain a query vector; inputting the query vector into a vector knowledge base, performing approximate nearest neighbor search on the query vector and a storage vector in the vector knowledge base, and inputting the storage vector meeting a preset semantic distance into a candidate text block sequence; analyzing the semantic association strength between each text block in the candidate text block sequence and the query vector; performing descending sorting on each text block in the candidate text block sequence according to the similarity score vector; and inputting the enhanced prompt text into the large language model for reasoning analysis to obtain a retrieval result. According to the method, the text semantic similarity and the context correlation can be considered at the same time in the retrieval process, and the accuracy of the retrieval result is improved through a dynamic weighting mechanism.

The embodiment of the invention provides a model optimization method, electronic equipment and a storage medium. The method comprises the following steps: acquiring an initial quantization bit width and an initial pruning rate of each convolutional layer in a BEV perception model; on the basis of the initialized quantization bit width of each convolution layer, quantizing the weight of the convolution layer, and according to the initial pruning rate of each convolution layer and the importance score vectors of the multiple output channels, obtaining pruning masks of the output channels; based on the loss function, the quantized weight and the pruning mask training model, obtaining an initial deployment model; and carrying out online quantitative sensitivity evaluation training on the initial deployment model, and when an evaluation condition is met, taking the initial deployment model as a final deployment model. Therefore, the problems that the domain adaptation and generalization ability is limited, the detection precision is reduced, the domain adaptation and generalization ability is difficult to be efficiently utilized by NPU of edge chips such as horizon lines, model evolution in the training process cannot be responded, and error accumulation is serious are solved, and structured sparsity, dynamic adaptation and end-to-end collaborative optimization are achieved.

The invention relates to the technical field of employee evaluation, in particular to a high-risk industry multi-dimensional dynamic weight employee evaluation method, which comprises the following steps of: obtaining quantitative scores of three-level indexes of employees and normalizing the quantitative scores by constructing a multi-level evaluation index system; inputting the standardized score vector into a Transform-based weight learning model, dynamically learning an index weight through a self-attention mechanism, carrying out step-by-step aggregation through a multi-layer perceptron, and outputting a comprehensive score of the employee on each first-level index; inputting the comprehensive score and the historical background information into a pre-established large language model, and performing fusion analysis according to a structured template to generate a personalized evaluation text; and finally, based on the score, the text and the employee information, automatically synthesizing a visual comprehensive evaluation report containing the radar map, the information bar and the text. According to the invention, objectiveness, individuation and operability of employee evaluation are realized.

The invention discloses an action sports scoring method and device based on big data analysis, and relates to the technical field of computer vision, and the method comprises the following steps: S1, constructing a human body posture tensor manifold; s2, generating aligned action track characteristics; s3, generating symmetrical positive definite manifold features; s4, generating deep manifold distribution parameters; s5, quantifying global distribution deviation characteristics; s6, calculating a residual vector based on the deep manifold distribution parameters and the standard action distribution parameters, inputting the residual vector into the improved AGCN model for processing, constructing an adaptive topology based on the residual vector, and performing aggregation analysis to obtain a joint-level physical angle error after multi-scale space-time convolution and manifold enhancement; and S7, outputting a comprehensive score vector. According to the method, the limitation that a traditional method only depends on local geometric features and ignores global topological structure constraints and statistical distribution priori is overcome, and an efficient solution is provided for intelligent scoring of sports actions.

The invention relates to a multi-causal model integrated abnormal root cause tracing method and device and computer equipment. The method is applied to industrial equipment, and comprises the following steps: acquiring multivariable time sequence data of the industrial equipment in an abnormal state; based on the multivariable time series data, training at least three causal inference models adopting different prediction functions; determining a unified root cause score vector based on the root cause score vector of each causal inference model after training; and determining a root dependent variable based on the unified root cause score vector. By adopting the method, the root cause identification efficiency and accuracy can be improved.

The invention provides a personnel portrait construction method and system based on a multi-dimensional talent evaluation model, and belongs to the technical field of human resource management and data processing, and the method comprises the steps: collecting the original data of talents, constructing a tree-level comprehensive quality index system, and outputting the index score vectors of the talents; performing dimension reduction processing on the original data to extract key features, calculating objective weights of the key features, and outputting weight vectors; inputting the key features and the weight vectors into a random forest machine learning model to calculate a preliminary evaluation score; on the basis of the key features, an analytic hierarchy process is adopted to calculate expert correction scores; integrating the two scores to calculate a comprehensive evaluation score, and constructing a personal portrait based on the index score vector, the comprehensive evaluation score and the key features; and updating the personal portrait, monitoring the index score vector or the comprehensive evaluation score, and generating growth early warning information for the talents meeting the early warning conditions. According to the method, multi-dimensional data are integrated, the weight is objectively calculated, and the portrait is dynamically updated.

The application discloses a kind of based on timing characteristic analysis one section converter operating abnormality detection method, to be in by analyzing the timing characteristic of one section converter sampling data, complete the real-time abnormality detection of one section converter operating state.Specifically, the method of the present application carries out timing characteristic analysis on multiple time nodes for the sample data collected when one section converter is in normal operation, establishes data model for online real-time abnormality detection after learning corresponding timing characteristic, and then detects whether the operation of one section converter is abnormal through a comprehensive abnormality detection index.Compared with prior art, the method of the present application obtains corresponding feature transformation matrix through timing characteristic analysis process capable of minimizing the difference between score vectors in time sequence, and completes the real-time online abnormality detection of one section converter on the basis of monitoring through the comprehensive abnormality detection index of three different abnormality detection indexes.

The invention discloses an employment psychological analysis model construction method and an employment psychological analysis method, and belongs to the field of artificial intelligence, and the method comprises the steps: distributing a corresponding dynamic distribution weight for each psychological traits dimension according to a center vector of each psychological traits dimension and an original embedded vector of each word, combining the projection matrix and the original embedded vector of each word to generate a characteristic projection embedded vector of each word; calculating an intensity score vector of each psychological traits dimension according to the traits projection embedding vector of each word in combination with high and low expression intensity reference vectors, and adjusting a universal language model in combination with a position coding vector, the traits projection embedding vector and an attention head output vector; and training the adjusted model by using the employment psychological assessment question and answer data set to obtain an employment psychological analysis model. By implementing the application, the problem that a general language model in the prior art cannot decouple the employment psychological dimensions of the college students and cannot capture the employment psychological traits of the college students can be solved.

The invention relates to the technical field of information security, in particular to a penetration test effect evaluation method and system based on an attack mode framework, and the method comprises the steps: firstly, constructing an evaluation model comprising a target layer, a criterion layer and an index layer; the evaluation indexes under each criterion are weighted through a large language model, and the indexes are quantitatively scored by experts. And for each criterion, on the basis of a grey theory, calculating a comprehensive evaluation vector of the criterion by using index scores, weights, a preset comment set and a whitening weight function so as to obtain a quantitative score and a qualitative evaluation result. And finally, calculating a final comprehensive evaluation result of the target layer by integrating the quantitative score vectors and the criterion weights of all the criteria. During decision making, the final result needs to be combined with the qualitative evaluation conclusion of each criterion so as to formulate a comprehensive and targeted security reinforcement strategy. According to the invention, the accuracy of penetration test effect evaluation is improved.

The invention provides a multi-agent collaborative illusion suppression method and device, equipment, a medium and a product, and the method comprises the steps: carrying out the multi-label classification of a user request through a dominant large language model, and generating a request classification weight vector; then, combining the ability score vector of each cooperative large language model to calculate the adaptation degree with the current request, and screening out a plurality of target cooperative models according to the adaptation degree; after the target models process the same request in parallel and generate responses, the dominant model carries out modular division on the responses, and for each module, weighted fusion is carried out on output content of the module according to the adaptation degree of the target models to obtain final response content of the module; and finally, combining fusion results of all the modules according to an original module sequence to form a unified collaborative response. In conclusion, the stability, the response efficiency and the generalization ability of the large model on illusion suppression are effectively improved.

This invention belongs to the field of data processing, specifically disclosing a clustering method, system, terminal, and storage medium for mass spectrometry data. The method includes: acquiring a target mass spectrometry dataset, performing state freezing processing and affinity calculation to obtain an affinity score vector; determining a preset similarity selection strategy, and partitioning the target mass spectrometry dataset according to the preset similarity selection strategy and affinity score vector to obtain the optimal assigned cluster; updating the centroid of the optimal assigned cluster to obtain the updated centroid, and performing deterministic structure editing processing on the optimal assigned cluster to obtain the target assigned cluster; constructing a kinetic potential function, and performing iterative clustering processing on the target assigned cluster to obtain the final clustering result. This invention, by performing state freezing processing on mass spectrometry data, enables parallel processing of mass spectrometry data, effectively improving the efficiency of mass spectrometry data clustering analysis and the accuracy of clustering result output.

The application provides a relationship extraction method and system based on entity replacement and floating mark, and relates to the field of natural language processing.The method comprises the following steps: performing entity replacement and floating mark processing on the original text to obtain a detection text; performing language feature extraction processing on the detection text to obtain a feature vector; performing vector splicing processing and language feature extraction processing on the feature vector to obtain an entity relationship score vector; performing probability distribution prediction processing on the entity relationship score vector to obtain a category probability distribution graph; and performing decoding processing on the category probability distribution graph.The entity pair of the relationship to be extracted is replaced with a fixed mark in the application, so that the expression of the relationship is more general, and the relationship classification error caused by the name interference of the entity pair is solved; meanwhile, special representations are added before and after the entity pair, and the floating mark is marked at the end of the sentence, so that the problem of poor model generalization performance caused by insufficient representation of entity information in the relationship extraction process is solved.

The invention discloses an intelligent gateway multi-protocol routing optimization method based on reinforcement learning. The method comprises the following steps: constructing a gateway observation state vector containing a multi-protocol interface link quality parameter, a queue state parameter and a forwarding consistency parameter in a preset sampling period; generating a routing credibility score vector by adopting a weighted score deduction strategy, and splicing the routing credibility score vector with the gateway observation state vector to obtain an adversarial perception state vector; inputting the adversarial perception state vector into a route optimization model based on a deep Q network to obtain a candidate route action set, and performing action screening based on a route credibility score vector; in the execution stage, protocol switching is triggered according to the condition that the credibility of the current protocol is lower than a switching threshold value and the credibility of the alternative protocol is higher than an access threshold value, and the next hop with the minimum end-to-end hop count is selected to complete forwarding during triggering. Therefore, the misleading probability of the abnormal measurement to the learning decision is reduced, and the continuity and the security robustness of the industrial Internet of Things service are improved.

The application discloses a malicious information detection method and device, electronic equipment and storage medium. The method comprises the following steps: obtaining original data based on preset acquisition conditions; preprocessing the original data to obtain a score matrix; determining target features based on the score matrix; weighting and aggregating the target features, filtering the original data according to the weighted and aggregated feature values to obtain an original data set; determining a target user according to the original data set, dividing the score data set of the original data into a target data set of preset data through a random discarding mode, and then performing implicit semantic processing on each data set to obtain a score vector of the target user; determining the distance difference between each target data set based on the score vector of the target user; and then determining the detection result of the malicious information of each target data set according to the distance difference. The malicious information detection can be efficiently and accurately realized, and the application can be widely used in the technical field of computers.

The invention relates to the technical field of psychological assessment, and discloses a multi-modal psychological feature modeling and assessment method based on artificial intelligence, and the method comprises the steps: carrying out the preprocessing of a house tree person drawing image, and extracting image features from the preprocessed house tree person drawing image; a score vector model is generated, image features are converted into structured vectors, a Sigmoid function is optimized and trained, scores of different psychological dimensions are calculated through the Sigmoid function, accurate quantitative evaluation can be provided for the psychological state, emotional tendency and the like of an individual, meanwhile, an SHAP method is used for interpretable analysis, the prediction result of the model can be analyzed, and the prediction accuracy of the model is improved. And the contribution of each feature to the result is quantified, so that the transparency of the model is enhanced, a tool for understanding the model decision process is provided for a user, the credibility of the model is improved, and the psychological feature evaluation accuracy is improved.

The invention discloses a softmax calculation method and device for an image classification model, and relates to the technical field of neural networks, and the method comprises the steps: obtaining a classification score vector, obtaining a classification score vector after numerical stabilization processing, carrying out the descending sorting, and obtaining a sorted classification score vector; respectively calculating an approximate index value of each element in the classification score vector, and respectively calculating a contribution degree corresponding to each element; screening according to the values of the contribution degrees to obtain a reserved item set and a perforated item set; respectively calculating the approximate normalization probability of each element, and obtaining a probability vector based on the approximate normalization probability of each element; according to the method, only the reserved items with remarkable contribution degree are subjected to approximate operation, and the perforated item elements with extremely low contribution degree are subjected to simplified processing, so that the number of elements needing to execute index operation is further greatly reduced, and the consumption of a large amount of index calculation is avoided.

This invention discloses an inference method and system for gastrointestinal endoscopic images based on a multimodal large model. The method includes: removing redundant background from the gastrointestinal endoscopic image input to the visual encoder in the multimodal large model based on the sparsity of feature distribution; calculating the local probability density, self-information, and global dependency score of the visual token in the gastrointestinal endoscopic image; calculating the retained score vector of the visual token by weighting the self-information and global dependency score; performing a first pruning process on the gastrointestinal endoscopic image based on the retained score vector; and inputting the pruned gastrointestinal endoscopic image and text information into the decoder in the multimodal large model for decoding to obtain the inference result of the gastrointestinal endoscopic image. This invention improves inference performance by pruning the gastrointestinal endoscopic image and extracting effective information before inference.

The invention relates to a response method and device based on charging intelligent customer service dialogue state tracking, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring a predefined intention slot structure definition set and a historical dialogue data set, and inputting the intention slot structure definition set and the historical dialogue data set into a graph neural network model to generate a state graph structure; obtaining semantic input content of a user in a current round, generating a node context sensing embedding set and a node semantic response intensity set in combination with the initial node embedding set, and determining candidate state scores in combination with a platform service state set and a node frequency scalar set; generating candidate state sub-graphs and candidate state score vectors according to the candidate state scores; and executing graph neural network propagation on the candidate state sub-graph, and determining a current dialogue state, a corresponding slot set and a state propagation path to generate a response strategy for responding to the semantic input content. By adopting the method, the accuracy of the response to the user request can be improved.

The invention discloses a cloud multi-modal intelligent recommendation method and system for consumption decision making, and relates to the technical field of artificial intelligence and recommendation systems. The method comprises the following steps: acquiring image, voice, text and behavior data in a user interaction process, and constructing a unified multi-modal input structure; behavior preference, context and time dynamic characteristics are extracted based on the structure, and multi-dimensional intention representation is generated; performing cross-modal semantic association and structure enhancement in combination with commodity attribute embedding, outputting semantic fusion features, constructing a correlation retrieval structure, and generating an initial candidate commodity set; in combination with the user behavior sequence, consumption tendency is evaluated, behaviors are matched, and recommended candidate score vectors are generated; and finally, optimizing and generating a final recommended commodity sequence in combination with the consumption constraint parameter, and synchronously outputting the final recommended commodity sequence to a user terminal. Through multi-modal deep fusion and dynamic intention modeling, in combination with cross-modal semantic enhancement and constraint optimization mechanisms, accurate and efficient personalized consumption recommendation meeting actual requirements is realized.

This application provides a vehicle collision recognition method and a vehicle, capable of jointly extracting features from acquired raw acoustic and vibration signals to obtain a topological embedding feature vector; performing temporal causal modeling based on the topological embedding feature vector to obtain causal information; and performing multi-agent collaborative discrimination based on the causal information and the topological embedding feature vector to obtain an anomaly score vector; performing multi-source information fusion decision-making based on the anomaly score vector, causal information, and the topological embedding feature vector to obtain fusion evaluation data; determining a dynamic alarm threshold based on the real-time vehicle status; and determining the collision recognition result based on the dynamic alarm threshold and the fusion evaluation data. By introducing a joint modeling mechanism for acoustic and vibration signals, this method addresses the current problem of insensitivity to low-amplitude undercarriage impacts on vehicles.

The invention discloses a method for automatically sensing and aligning personalized preferences of a large language model, and relates to the field of natural language processing. According to the method, a triple data set of a user question text, a target answer text and a multi-dimensional real preference score vector is collected, and a double-head architecture model containing a shared backbone network, an adaptive preference perception module and a condition generation module is constructed; constructing a total loss function based on weighted multi-dimensional mean square error loss and sequence-level cross entropy loss, and performing end-to-end joint optimization on model parameters; during reasoning, a user only needs to input a natural language question, and the model automatically predicts a preference score vector and generates a personalized reply. According to the method, the problems of inconsistent training-reasoning, high user interaction burden and limited reasoning efficiency of a traditional method are solved, and accurate alignment and high-quality personalized generation of multi-target preferences are realized while low resource consumption is ensured.

The application discloses a wind turbine generator unit variable flow system performance degradation identification method and related device, comprising: collecting the temperature of IGBT module in the variable flow system, the DC bus voltage, the output power harmonic distortion rate, and constructing a data set based on the same; performing data processing on the data set to obtain a processed data set; performing nonlinear feature extraction on the data set to obtain a dimension-reduced principal component score vector; constructing a hidden Markov model; and calculating the probability of the wind turbine generator unit variable flow system being in each degradation state based on the hidden Markov model and the dimension-reduced principal component score vector. The method and related device can realize performance degradation identification of the variable flow system, and have lower requirements on data quality.

The invention relates to the technical field of artificial intelligence, in particular to an application program performance optimization method and device, equipment, a medium and a product. The method comprises the steps that source codes of an application program are obtained, and a plurality of optimization strategies are determined; obtaining a static optimization factor based on the optimization effect of the optimization strategy; analyzing the running log of the application program to obtain corresponding system index information; comparing the system index information with a preset index to obtain a dynamic optimization factor; utilizing a deep learning model to confirm a performance score vector of the application program; and determining a performance optimization scheme based on the performance score vector. The static performance of the application program can be analyzed in real time by performing exception analysis on the source code; by analyzing the running log during running of the application program, the suspicious behavior during running of the application program and the influence degree on the performance can be analyzed, the comprehensive performance conditions of program code development and real-time running are considered, and the method is more flexible and intelligent.

This invention discloses an intelligent network attack attribution method based on federated learning, comprising: S1, collecting time-stamped network communication event sequences locally at each federated node to construct a dynamic graph structure; S2, extracting time-varying graph statistical feature sequences from the dynamic graph structure; S3, performing change point detection based on structural change cost and combined with minimum stage length constraints to generate a set of attack evolution stage boundaries; S4, mapping stage boundaries to stage identifier sequences and writing stage association labels for edges; S5, generating edge-level stage gating coefficients based on stage association labels; S6, introducing stage gating coefficients into the tangent space of a hyperbolic graph neural network to perform message passing and aggregation, generating node embedding vectors and edge embedding vectors; S7, forming attack source candidate scoring vectors locally at each federated node and performing aggregation calculations, outputting the attack source ranking results and propagation path representation. This invention achieves adaptive attribution of attack stages in a federated scenario.

The application provides an explainable coronary heart disease noninvasive screening method and system based on multi-agent decision, and the method comprises the following steps: encoding according to a preset field order based on noninvasive health assessment data to obtain an input vector, inputting the input vector into a grade score prediction model, outputting a score vector of each candidate risk grade, and selecting the grade corresponding to the maximum score vector as a target risk grade. Evidence description information is constructed according to the source information of the noninvasive health assessment data and the related information of the target risk grade, and the evidence description information is divided into multiple evidence units according to different data types. It is judged whether each evidence unit has the ability to support the target risk grade, and the evidence unit judgment result is obtained, and each evidence unit corresponds to a stability weight. The establishment judgment result is calculated according to the evidence unit judgment result, the stability weight and the establishment threshold, and it is judged whether the target risk grade is established at the evidence level according to the establishment judgment result, and the reliability of disease screening is improved.

The application discloses a convolutional decoder, a decoding method and an electronic device, relates to the technical field of communication, and is used for reducing time overhead of convolutional code decoding. The application is used for decoding an encoded sequence, the encoded sequence is divided into P first sub-sequences, tail sequences of the P first sub-sequences are P second sub-sequences, and the convolutional decoder comprises the following steps: for first round decoding, P Viterbi decoders decode the P second sub-sequences respectively, and update cumulative score vectors of the P second sub-sequences; from second round decoding, a pth Viterbi decoder starts to perform cyclic decoding on the encoded sequence from a starting position of a pth first sub-sequence, and update cumulative score vectors of the P first sub-sequences; P comparators compare the cumulative score vectors of two sub-sequences with the same ending position in the last two rounds of decoding, and if the maximum probability state transition paths of the two sub-sequences with the same ending position are the same, the decoding sequence of the maximum probability state transition path is output.

This application discloses a text readability analysis method, system, and electronic device based on NLP processing. The method extracts multiple character entities; performs word segmentation on the preprocessed text to be analyzed to obtain a word sequence and constructs a dependency syntax tree; based on the multiple character entities, the preprocessed text to be analyzed, and the dependency syntax tree, it extracts features such as the number of characters, the entropy of character personality diversity, the conflict density, the dialogue ratio, the relationship density, and the relationship complexity; it processes and fuses these features to obtain a fused feature vector; it inputs the fused feature vector into an encoder to obtain a preliminary score vector; and it determines the readability score of the text to be analyzed based on the preliminary score vector. This application can improve the accuracy of text readability analysis.

The application provides a kind of commuting network evaluation method and system based on artificial intelligence, through S1 generate the smooth manifold of commuting network on Lie group space, according to S2 construct dynamic traffic manifold, through S3 construct multi-agent forest model, generate commuting score vector through agent forest model, then through S4 generate score relationship equation to integrate commuting score vector, and the three commuting paths of highest score are marked and pushed to the user, considering various commuting factors, so that the final commuting score is more accurate, while adding the historical data of the user, more in line with the needs of the user, through S5 obtain the final commuting path of the user and carry out tracing, while generating traceability report, according to S6 optimize dynamic traffic manifold, and return to S2, find the factors affecting the score by the method of tracing and analyze, while optimizing the dynamic traffic manifold according to the analysis result, to achieve the purpose of optimization evaluation.