533 results about "Sample Label" patented technology

Provided is a defect detection method and device, computer equipment and a storage medium. The method includes: acquiring an RGB image, a depth image and a sample label of a detection object sample; performing feature map extraction and feature map fusion on the RGB image and the depth image by a feature extraction network of the defect detection model, to obtain a fused feature map; performing defect detection based on the fused feature map by a feature reconstruction network of the defect detection model, to obtain a defect score map, wherein the defect score map being obtained by fusing a global defect score map which is generated based on a global defect detection network with a local defect score map which is generated by a local defect detection network; and updating parameters of the defect detection model based on the defect score map and the sample label.

The invention provides a training method of a document image tampering detection model and a tampering detection method and device.The training method of the document image tampering detection model comprises the steps that multi-scale visual domain features are extracted from a sample document image, and multi-scale frequency domain compressed sensing features are extracted from frequency domain information; acquiring tampered area edge mask data from the document image; fusing the multi-scale visual domain features and the multi-scale frequency domain compressed sensing features to obtain multi-modal fusion features; performing semi-supervised training on the multi-scale sensing network by taking the multi-scale visual domain feature as a sample feature of a first prediction head, taking the multi-modal fusion feature as a sample feature of a second prediction head, taking a real label or a pseudo label as a sample label and taking joint loss as a loss function to obtain a document image tampering detection model; according to the method provided by the invention, document image tampering pixel-level detection under low labeling cost is realized, and the detection precision of a document image tampering detection model is improved.

The invention belongs to the technical field of aero-engine state monitoring and intelligent fault diagnosis, and discloses a multi-modal information fusion bearing fault diagnosis method based on self-supervised learning. The method comprises the following steps: firstly, through mask reconstruction self-supervision pre-training, extracting stable feature representation insensitive to mask disturbance from an unlabeled multi-modal signal, and dynamically updating each modal feature reference point by using an index moving average algorithm; in a downstream fault diagnosis task, a multi-modal joint decision model comprising a pre-training encoder, a single-modal classifier and a fusion classifier is constructed, and adaptive weighted fusion of multi-modal decision is realized through contribution degree calculation based on a cooperative game Shapley value in combination with a deviation degree of modal features and a reference point. According to the method, the dependence of the deep neural network on fault labeling data is effectively reduced, the accuracy and robustness of the diagnosis system in a multi-modal signal diagnosis scene are improved through a dynamic fusion mechanism, and the method is suitable for industrial scenes with limited sample label resources.

The invention discloses a medical image recognition system based on label noise robust learning, and provides a hard sample label refinement strategy based on a confidence perception weighted prototype and an effective noise sample joint correction method to process divided subsets in correction before training so as to obtain training data with higher quality; in progressive hard sample reinforcement learning, data are input according to the learning difficulty of samples for training, the ability of model learning discrimination feature representation is improved by integrating cross entropy loss, consistency loss and credible contrast loss, and the influence of medical tag noise is reduced. According to the method, by integrating label refinement and a progressive hard sample reinforcement learning technology, the accuracy and robustness of medical image recognition under the condition of noise label data are effectively improved, overfitting of the model to noise samples is relieved, and the method can be used for correctly building a disease auxiliary diagnosis model under the condition that the noise label samples exist in training data.

The invention discloses a data processing method, device and equipment and a readable storage medium, and the method comprises the steps: inputting business sample data into an initial large language model, predicting the business sample data through the initial large language model, and obtaining M lexical element prediction vectors; based on the M lexical element prediction vectors, generating a sample text result and a sample confidence coefficient corresponding to the sample text result; generating a correctness reward value based on a sample label corresponding to the business sample data and the sample text result, and generating a confidence coefficient calibration reward value based on the sample confidence coefficient, the sample label and the sample text result; and based on the confidence calibration reward value and the correctness reward value, performing reinforcement learning training on the initial large language model to obtain a target large language model. By adopting the method, the confidence of model output and the accuracy of a prediction result can be improved.

The invention discloses an online incremental learning method and system based on a self-adaptive B + tree index, and relates to the technical field of artificial intelligence, and the method comprises the steps: constructing an extended self-adaptive B + tree index structure, inserting a new sample, and calculating a trajectory disturbance value; generating a disturbance score by combining the disturbance value and a node state, and adding a new sample and an associated node sample into an incremental learning scheduling queue when a condition is met; when a node splitting, merging or reconstructing event occurs, acquiring an affected node and updating a structural code of the affected node; monitoring sample label track change, adjusting a label confidence mark when a threshold value is exceeded, and limiting participation frequency according to a screening mechanism; and performing forgetting compression operation on the leaf nodes with low access frequency and small label fluctuation degree. By constructing an adaptive B + tree index and combining disturbance-driven scheduling, structure change response and forgetting compression strategies, sample management refinement, learning update high efficiency and structure maintenance controllability are realized, and the stability and resource utilization efficiency of online incremental learning are improved.

The invention relates to a tool wear compensation control method, device and equipment and a medium, and relates to the field of machining control, the method comprises the following steps: adopting a wear state label with a label to carry out supervised constraint, adopting a Baum-Welch algorithm with label correction to carry out iterative optimization, and adopting a Baum-Welch algorithm with label correction to carry out iterative optimization; enabling the wear detection model to output an optimized state transition probability matrix and an optimized observation probability matrix; the wear feature vectors of different time steps, the optimized state transition probability matrix and the optimized observation probability matrix serve as training samples, the tool compensation amount corresponding to the wear feature vectors serves as a sample label, and a wear compensation prediction model is trained to be in a convergence state; and inputting the wear feature vector of the current time step into a wear compensation prediction model to predict the tool compensation amount of the operation tool in the future time step, and inputting the tool compensation amount into the numerical control machine tool to carry out tool wear compensation. The wear compensation amount of the tool can be predicted in real time and compensated, so that the service life of the tool is prolonged.

The embodiment of the invention provides a model training method, a facial expression recognition method and device and electronic equipment, and relates to the technical field of video processing. The model training method comprises the following steps: acquiring a sample video and a first sample label; extracting a time feature and a space feature of the sample video by using a time-space feature extraction network in the facial expression recognition model of the initial structure; calculating an attention weight representing the correlation between the spatial feature and the time feature by using a mapping network; performing weighted aggregation on the time features by using the attention weight to obtain fused spatio-temporal features; inputting the fused spatial-temporal features into a classification network to obtain a first recognition result; and performing model training based on the difference between the first recognition result and the first sample label to obtain a trained facial expression recognition model with higher accuracy.

The invention relates to the technical field of data processing, in particular to the technical field of artificial intelligence, and provides a mining model training method and device, electronic equipment and a storage medium which are used for improving venation mining efficiency and quality. The method comprises the following steps: training an initial mining model based on a first sample set to obtain an intermediate mining model; the model corresponds to a plurality of tasks, and each task represents a sub-process of the event venation mining process; the first sample set comprises a first sample and a sample label associated with each task; inputting at least one second sample associated with the task into the intermediate mining model to obtain a candidate result set output for each second sample; screening each candidate result contained in each obtained candidate result set according to a preset evaluation standard to construct a second sample set; the second sample set comprises each second sample and the corresponding positive candidate result and negative candidate result; and training the intermediate mining model based on the second sample set to obtain a target mining model.

The invention relates to a water supply network leakage identification method and device based on data enhancement and hybrid neural network architecture, equipment and a medium, and the method comprises the steps: employing a generative adversarial network to generate a simulation pipeline vibration audio signal, and combining the simulation pipeline vibration audio signal with a real pipeline vibration audio to construct a pipeline vibration audio signal data set; performing feature fusion on the Mel-frequency cepstral coefficient feature matrix and a first-order difference matrix of Mel-frequency cepstral coefficients to determine a feature fusion matrix of each pipeline vibration audio signal frame; by taking the feature fusion matrix as a training sample and the leakage state of the pipeline vibration audio signal as a sample label, training a CNN-Bi LSTM hybrid neural network model to determine a water supply network leakage detection model; and inputting the to-be-identified pipeline vibration audio signal into the water supply network leakage detection model to determine whether the to-be-identified pipeline vibration audio signal is in a leakage state or a non-leakage state. According to the invention, the precision, robustness and generalization ability of leakage detection are significantly improved.

The invention discloses an agent training data set construction method and system combined with cross-modal learning, and relates to the technical field of agent training, and the method comprises the steps: obtaining multi-modal intelligence data from a multi-source database, extracting feature information, and projecting the feature information to a preset semantic space to obtain cross-modal alignment features; constructing an agent function call syntax tree based on cross-modal alignment features, analyzing the syntax tree into an instruction sequence, constructing an analysis reasoning chain, and generating a sample label and an interaction track; constructing a task dependency graph by using the sample labels and the interaction tracks, decomposing the task dependency graph into a plurality of parallel decision branches, and dynamically adjusting a processing strategy to form a training data set; mapping the training data set to an agent target space, optimizing and analyzing an inference chain through a training feedback channel, and outputting a standardized sample library; and finally inputting to-be-analyzed data into the trained agent, and generating an intelligence analysis report according to the analysis reasoning chain.

The invention relates to an intelligent agent continuous training and effect evaluation closed-loop method and system fusing work order feedback, and a medium, and relates to the technical field of deep learning. The agent continuous training and effect evaluation closed-loop method comprises the following steps: receiving and analyzing work order feedback information, and recording the number of error information and information storage duration in combination with a preset structure sample label; dynamically judging the number of error information and the information storage duration, triggering a retraining rule of an original agent model, distributing candidate information weights, and screening a target sample set; directionally training and optimizing the original agent model according to a retraining rule in combination with the target sample set, and generating a new agent model; constructing a historical work order problem set, comparing and verifying a new agent model in combination with an original agent model, generating an agent evaluation table, and correcting a candidate information weight; by accurately focusing the weak link, the continuous and efficient enhancement of the model capability is realized, so that the answer accuracy of the intelligent agent is remarkably improved, and the AI illusion is effectively inhibited.

Provided in the present application are a model training method and apparatus, a speech detection method and apparatus, and a device, a medium and a product. The model training method comprises: acquiring at least one positive example speech sample, at least one noise sample, and at least one negative example interference sample comprising speech, wherein a sample label of the positive example speech sample represents a positive example sample, and sample labels of the noise sample and the negative example interference sample represent negative example samples; obtaining sample detection results of a speech detection model performing speech detection on the at least one positive example speech sample, the at least one noise sample and the at least one negative example interference sample, wherein the sample detection result of a sample represents whether the sample belongs to the positive example sample; on the basis of the sample detection results and the sample labels, determining a total training loss, and on the basis of the total training loss, performing iterative training on the speech detection model, so as to obtain a trained speech detection model.

The invention relates to the technical field of pattern design intelligent optimization, in particular to an embroidery pattern optimization design method and system based on a machine learning model.The method comprises the following steps that a grey-scale map structure is obtained, edges are extracted to construct a direction node map, node intersection points are positioned to generate path vectors, boundaries are drawn according to the symmetrical relation to generate annotation map blocks, and the annotation map blocks are obtained. Establishing a pattern code generation structure label group, and extracting a repeated paragraph classification style to generate a style training sample label set; according to the method, a composition node map is constructed through the direction continuity of edge pixel points, path candidate intersection points are positioned through the intensity of node space distribution and are connected into skeleton path line segments, an edge structure closed area is extracted by using a path symmetric projection relation, and ordered association of a pattern structure and a pattern is realized. The pattern design process has the capabilities of being controllable in structure, adjustable in style and traceable in feature, and the intelligent level of composition processing and style induction is remarkably improved.

The invention provides an incremental learning method for open world target detection based on a large language model. The method comprises the steps that a first RGB image sample marked with a first type of target is acquired and trained to obtain a first open world target detection model; establishing a training set comprising a plurality of second RGB image samples, wherein each second RGB image sample is labeled with targets of a first category and a second category; processing the first category and the second category by utilizing a large language model, and generating an attribute feature in a text form of each category; processing the second RGB image sample and the text-form attribute features of each category by using a first open world target detection model to obtain a target frame prediction value, a target category prediction value and an unknown category target prediction value, thereby determining a total loss value; and updating parameters of the first open world target detection model based on the total loss value, thereby obtaining a second open world target detection model. The generalization ability and adaptability of the model in a new scene are enhanced.

The invention relates to the technical field of crop remote sensing samples, and discloses a crop remote sensing image intelligent sample library construction method based on a knowledge graph. The method comprises the steps of collecting multi-source remote sensing image data, and generating an enhanced image data set through fusion and enhancement processing; and constructing a crop field knowledge graph integrating the crop growth model, the soil type and the climate condition. A knowledge graph is used as guidance, crop areas are recognized through semantic analysis, multi-scale features are extracted, and category attributes and associated contexts are deduced through a graph reasoning algorithm in combination with a relation path; and generating a sample label candidate set through deep learning of the graph attention network, screening and correcting the sample label candidate set by using an optimization algorithm constrained by the knowledge graph, and outputting high-quality sample labels. And integrating and storing labels and image data into a graph structure sample library, implementing a dynamic updating mechanism to adapt to new data, running a quality monitoring process driven by a knowledge graph, and adjusting a construction strategy according to result feedback to realize intelligent and efficient construction of the sample library.

The invention discloses a magnetic data boundary identification method and system based on a physical prior selection module, and relates to the technical field of geophysical exploration, and the method comprises the steps: building a magnetic anomaly data set through underground target magnetic anomaly data with different parameters; generating magnetic measurement data through a magnetic anomaly forward modeling calculation formula, and forming a magnetic measurement data set; constructing a boundary recognition network model based on UNet + +, and integrating the physical prior selection module to the boundary recognition network model; inputting the training set sample, the verification set sample and the test set sample into a physical prior selection module for operator transformation, and generating a corresponding transformation result graph; and calculating the structural similarity index of each transformation result graph and the corresponding sample label, and splicing the transformation result graph with the maximum structural similarity index and the corresponding sample according to the channel dimension to generate a corresponding multi-channel feature. According to the method, the boundary recognition network model is established and the physical prior selection module is added, so that the robustness of the boundary recognition network model to complex noise is improved.

The invention discloses a micro-plastic type and aging degree combined identification model training and application method and a related device, and relates to the technical field of micro-plastic identification, and the method comprises the steps: obtaining a sample hyperspectral image and a sample label of each micro-plastic sample in a plurality of micro-plastic samples, the sample label comprises a micro-plastic type and a micro-plastic aging degree, extracting a sample spectrum feature vector of each sample pixel point in the sample hyperspectral image, setting the sample label corresponding to the sample hyperspectral image as the sample label of each sample pixel point in the sample hyperspectral image, constructing a data set, and extracting the sample spectrum feature vector of each sample pixel point in the sample hyperspectral image; the data set comprises a sample spectrum feature vector and a sample label of each sample pixel point, constructing a deep learning model, training the deep learning model by using the data set, and obtaining a micro-plastic type and aging degree combined identification model. And the identification efficiency and accuracy are improved.

The invention relates to a frequency hopping signal detection method and system based on time-frequency compressed sampling deep learning, belongs to the field of signal processing, and solves the problems of low detection probability and high calculation complexity of a frequency hopping signal under a low signal-to-noise ratio. Comprising the following steps: respectively acquiring a frequency hopping signal when the frequency hopping signal exists and noise data when the frequency hopping signal does not exist, and carrying out time-frequency characterization and compressed sampling processing to obtain a frequency hopping time-frequency graph and a noise time-frequency graph; constructing a sample data set based on the frequency hopping time-frequency graph, the noise time-frequency graph and corresponding sample labels; constructing a frequency hopping signal detection network based on a lightweight convolutional neural network ShuffleNet and a convolutional attention module CBAM, and obtaining a trained frequency hopping signal detection network by using the sample data set; acquiring a time-frequency graph corresponding to a to-be-detected signal, and inputting the time-frequency graph into the trained frequency hopping signal detection network to obtain a confidence coefficient of the to-be-detected signal; and performing judgment based on the confidence of the to-be-detected signal to obtain a detection result of the to-be-detected signal. The detection performance is improved; and the reasoning complexity is reduced.

The application relates to a water supply network leakage identification method and device based on data enhancement and a hybrid neural network architecture, equipment and a medium. The method comprises the following steps: generating a simulated pipeline vibration audio signal by using a generative adversarial network, combining the simulated pipeline vibration audio signal with a real pipeline vibration audio signal, and constructing a pipeline vibration audio signal dataset; fusing a mel-frequency cepstral coefficient feature matrix and a first-order difference matrix of the mel-frequency cepstral coefficient to determine a feature fusion matrix of each pipeline vibration audio signal frame; taking the feature fusion matrix as a training sample, taking a leakage state of the pipeline vibration audio signal as a sample label, training a CNN-BiLSTM hybrid neural network model, and determining a water supply network leakage detection model; and inputting a to-be-identified pipeline vibration audio signal into the water supply network leakage detection model to determine whether the to-be-identified pipeline vibration audio signal is in a leakage state or a non-leakage state. The application significantly improves the precision, robustness and generalization ability of leakage detection.

The invention provides an aero-engine test process data processing method and system, and belongs to the technical field of data processing, and the method comprises the steps: training a deep learning model based on sample test process data and a corresponding sample label, and obtaining a test process data processing model; the method comprises the following steps: acquiring multichannel original data in an aero-engine test process, and preprocessing to obtain preprocessed test process data; identifying the preprocessed test process data by adopting a test process data processing model to obtain an engine test data identification result; the test process data, the recognition result, the test item information corresponding to the recognition result and the test item progress form structured data and are stored in blocks; when the structured data is accessed, decryption is carried out, and a visual data model is generated and displayed. According to the invention, the data processing efficiency, the data utilization efficiency, the test efficiency and the data analysis capability in the test process of the aero-engine are improved.

The invention provides a training method of a depressive emotion recognition model and a depressive emotion recognition method, and relates to the technical field of deep learning. According to the specific implementation scheme, sample video stream data and sample labels are obtained; obtaining a comparison description text generated by the large language model based on a preset cue word template; inputting the sample video stream data and the contrast description text into a neural network model to be trained, and obtaining an emotion recognition result output by the neural network model; and based on the emotion recognition result and the sample tag, training the neural network model to obtain a depressive emotion recognition model. According to the technical scheme provided by the invention, the recognition capability of the large language model on the depression emotion can be used as priori knowledge to be input into the neural network model in a form of generating the description text, so that the capture capability of the neural network model on the depression emotion related characteristics in the video stream data is improved; therefore, the emotion recognition effect of the model on low-quality multi-modal data is improved.

The invention relates to the technical field of remote sensing information, and discloses a corn fine classification method and system based on multi-source time sequence remote sensing image feature fusion, and the method comprises the steps: obtaining a multi-source remote sensing image and digital elevation model data of a target region; arranging corn and non-corn sample points in the target area, calculating multi-temporal vegetation indexes, texture features and gradient features based on the preprocessed remote sensing image, and constructing time sequence features; carrying out feature selection on the constructed time sequence features by utilizing importance screening and high-correlation feature redundancy elimination to obtain a feature subset; fusing the final feature subset with the gradient features to form a final optimized feature vector of each sample point; and adopting a Boosting ensemble learning algorithm, taking the optimized feature vector and the sample label as input, training a corn classification model, and predicting classification. The method has the advantage that high-precision automatic identification and classification of corn crops in a large-scale farmland can be realized.

The application discloses a large language model prompt word automatic optimization method, and relates to the technical field of prompt word optimization. The method comprises the following steps: setting an initial prompt word; constructing training data, a gradient generation prompt word template, a prompt word editing template, an optimizer and a task model; randomly sampling small batch data from the training data as a training set, and predicting an answer of the training set by the task model according to the initial prompt word; comparing the answer with sample labels of the training set to obtain an error example set; inputting the error example set and the initial prompt word into the gradient generation prompt word template to generate a gradient analysis request; receiving and analyzing the gradient analysis request by the optimizer model to generate a second natural language gradient; and inputting the second natural language gradient and the initial prompt word into the prompt word editing template, and modifying the initial prompt word by the optimizer model according to the editing prompt word template to generate a second candidate prompt word. The application solves the technical problem of low efficiency of manually writing prompt words.

The invention discloses an earthquake monitoring and early warning method and system based on an earthquake precursor, and relates to the technical field of earthquake forecasting, and the method comprises the steps: collecting the core data and auxiliary state data of the earthquake precursor, marking a sample label, and forming a training data set; establishing a core parameter screening model, and training the core parameter screening model by using the training data set to obtain core parameters; constructing a typical working condition library, calculating an Euclidean distance between a real-time data point and a historical working condition sample point, selecting first K neighbor points, and returning a working condition category with the highest frequency as a current operation working condition; corresponding anomaly prediction models are established according to the different operation conditions, and anomaly types under the different current operation conditions are predicted according to the anomaly prediction models; an EWI index is calculated through the core parameters, and when the EWI index is larger than an index threshold value, earthquake early warning is triggered; otherwise, continuously monitoring. According to the invention, real-time early warning of the earthquake precursor is realized under the condition that the equipment is abnormal.

The embodiment of the invention discloses a data processing method and device, equipment and a computer readable storage medium. The method comprises the steps of obtaining a training sample comprising a sample question, at least two sample answers and a preference tag; determining the at least two sample answers as sample tags of the sample question, and generating at least two supervised training samples according to the sample question and the at least two sample tags; adjusting parameters in the initial question and answer model according to the at least two supervised training samples to obtain a first question and answer model; inputting the training sample into a first question and answer model, and generating first prediction probabilities corresponding to the at least two sample answers in the first question and answer model; according to the preference label and the at least two first prediction probabilities, parameters in the first question and answer model are adjusted, and a second question and answer model is obtained. By adopting the method, the model training stability and the generation diversity can be improved.

The invention discloses an electricity larceny behavior detection method and system for an intelligent electric meter. The method comprises the following steps: collecting and preprocessing operation data of a target intelligent electric meter; performing feature extraction to generate a feature sample set; constructing a user baseline model, and outputting a reference feature vector; inputting the training sample into the improved RSAN anomaly detection model for training to obtain a trained improved RSAN anomaly detection model; generating a preliminary detection result; performing secondary fusion judgment to generate a final detection result; when the final detection result is an electricity stealing event, verification feedback data is formed; the sample labels of the training samples are corrected, an updated improved RSAN anomaly detection model is obtained, automatic detection and dynamic self-learning updating of the electricity stealing behavior of the intelligent electric meter are achieved, the detection precision is effectively improved, the false alarm rate is reduced, and the real-time response and self-adaptive optimization capacity is achieved.

The invention relates to a noise tag-based multi-mode identification method, device and equipment, a storage medium and a program product. The method comprises the following steps: acquiring a noise sample data set of a visual language model, wherein the noise sample data set comprises a plurality of image samples and sample tags corresponding to the plurality of image samples respectively; according to the semantic trust boundary of the sample labels, dividing the sample labels into trust labels and non-trust labels; according to the image sample of the trust label and the image sample of the non-trust label, prompt learning is carried out on the visual language model so as to optimize the visual language model; and identifying a to-be-identified image in a downstream task by using the optimized visual language model, and determining the category of the to-be-identified image. By adopting the method, the recognition accuracy can be improved.

The invention discloses a method for constructing a secondary subclinical hypothyroidism screening model based on conventional physical examination data, which comprises the following steps of: constructing a high-quality data set containing the conventional physical examination data, and defining a sample tag in the high-quality data set as a G2SCH positive group or negative group; performing feature selection on the data set by adopting different feature selection methods, and screening out key feature subsets; an oversampling method is adopted to synthesize minority class samples in the data set according to different class balance proportions; adopting a plurality of machine learning algorithms to carry out model prediction training with sample labels as supervision on the basis of data sets of different classes of balance proportions and different key feature subsets, and constructing a plurality of candidate G2SCH screening models; and evaluating and selecting an optimal G2SCH screening model. According to the method, G2SCH screening can be carried out based on conventional physical examination data, dependence on thyroid hormone specialized detection indexes in an existing medical scene is overcome, and the method is suitable for the medical scene with limited resources.

The invention discloses a multi-modal large model training method for a certificate OCR (optical character recognition) task. The method comprises the following steps: performing fine adjustment on a multi-modal large model through a small number of labeled samples; carrying out reinforcement learning training on the multi-modal large model through the remaining labeled samples until the recognition precision of the multi-modal large model on the verification set reaches a set standard, and completing the training of the multi-modal large model; the label of the sample label is structured output of the key field in the certificate image. According to the method, a long text task is disassembled into field-level feedback and optimization, relative advantages are generated in combination with intra-group comparison, so that effective learning in a sparse environment is realized, and the method greatly improves the recognition accuracy of a model to complex certificate contents while ensuring the normalization of an output format, and has a good application prospect. The method has the advantages that the method is simple and easy to implement, good generalization ability and reasoning level are shown, the multi-modal large model with excellent performance for the certificate OCR task can be trained under the condition that the manual labeling workload is reduced, and the method has wide application prospects.