15 results about "Model sample" patented technology

The application discloses a market supervision internet information monitoring and analysis system, which comprises a CNN model construction module, an internet information collection module, an information preprocessing module and an information mining module; the application uses a CNN model algorithm to model sample data marked by artificial marking, and uses a model to determine illegal behaviors; through continuously accumulating sample data, repeatedly training and achieving the purpose of machine learning, the determination accuracy is gradually improved.

This invention discloses a near-infrared model maintenance method, comprising: acquiring the pure spectral signals of each modeling sample of the near-infrared model to be maintained and the pure spectral signals of each test sample in the test sample set; performing principal component analysis on the pure spectral signals of each modeling sample to determine the principal component score and principal component model of each modeling sample; projecting the pure spectral signals of each test sample onto the principal component model to obtain the feature score of each test sample; determining each maintenance sample based on the principal component score and the feature score of each test sample in the test sample set; and updating the near-infrared model to be maintained based on each maintenance sample. This method can achieve rapid and accurate correction of the original model, not only expanding the coverage of its sample spectral principal component space but also expanding the applicability of the model, thereby making the model more adaptable to the prediction of new samples, and the prediction ability of the corrected model is greatly improved.

The application discloses an alumina substance content detection device and method, and belongs to the technical field of alumina substance content detection. The method of the application is as follows: 1, the alumina substance content detection device is used to configure the spectrum collection on the surface of a sample; 2, a spectrum matrix M of an nth sample is constructed n ; 3, the spectrum data obtained by the alumina sample is preprocessed; 4, a prediction model for detecting the alumina substance content of the in-model sample composed of 75% training spectrum data and 25% training spectrum data is constructed by a PCA-PLS linear regression sub-model and a preliminary model, the prediction model for detecting the alumina substance content is compared and verified by using out-of-model samples, and a trained prediction model for detecting the alumina substance content is obtained; 5, test data is input into the trained prediction model for detecting the alumina substance content, and a prediction result of the alumina substance content is obtained; compared with the prior art, the application improves the efficiency of the existing alumina detection.

The application provides a data privacy protection method and system based on adaptive weight adjustment, solves the problems of model performance decline and slow convergence speed caused by non-independent and identically distributed data, and belongs to the technical field of federated learning application. The method comprises the following steps: at the beginning of each round of federated communication, the server side evaluates the credibility of the global model at the category level by using an auxiliary data set, and downloads the credibility matrix and the global model parameters to the clients participating in the round of federation; the client evaluates the credibility of the global model at the sample level according to the local private data set, and uses the category credibility and the sample credibility for weighting when performing knowledge distillation, dynamically guides the training process of the local model, and uploads the updated local model parameters to the server side; and the server side aggregates the local model parameters to update the global model.

This application relates to the field of audit risk monitoring and data processing technology, and in particular to a method and system for audit risk monitoring, early warning, and prediction of energy enterprises. The method includes: acquiring relevant data from multi-source heterogeneous audits of energy enterprises; performing modeling sample screening and feature vector construction on the preprocessed relevant data, traversing the combinations of early warning units and audit risk indicators corresponding to the relevant data for screening, and obtaining the screened effective samples; processing the effective samples using a sliding window method to generate modeling data; training a pre-set CART regression tree prediction model based on the generated modeling data to obtain an audit risk indicator prediction model; acquiring the audit data to be monitored from the energy enterprise, obtaining the corresponding feature vectors to be predicted, inputting the feature vectors to be predicted into the trained audit risk indicator prediction model, and outputting the individual and combined predicted values ​​of the corresponding indicators for the month to be monitored, forming the audit risk early warning and prediction results.

A method, apparatus, non-transitory computer readable medium, and system for performing an accelerated denoising process includes obtaining a noise map from a noise distribution. Embodiments then compute, using an image generation model, a denoising vector based on an accelerated denoising trajectory, where the accelerated denoising trajectory accelerates a denoising rate based on a diffusion timestep. Then, embodiments generate, using the image generation model, a synthetic image by denoising the noise map based on the denoising vector.

The application provides an emulsion concentration detection method fusing impedance spectrum and PINN, comprising the following steps: collecting impedance modulus, phase angle and temperature data of the emulsion under multi-frequency excitation, calculating real and imaginary impedance and fitting equivalent circuit parameters representing electrochemical characteristics; fusing the equivalent circuit parameters and the temperature data to construct an electrochemical feature vector, and generating model samples through standardization processing; subsequently, constructing an emulsion concentration inversion model containing a bidirectional long short-term memory network and a physical constraint embedding module, the former extracts sample time sequence characteristics, and the latter constrains the physical relationship between the concentration and the equivalent circuit parameters; after the model is trained and optimized by using the model samples, the real-time collected test data is input, and the concentration detection result can be output; the application can effectively improve the precision, robustness and long-term stability of the online detection of the emulsion concentration under complex working conditions.

This application provides a rule and model generation method, apparatus, device, storage medium, and program product. The method includes: determining a feature set based on multidimensional heterogeneous data and a modeling sample list; generating an initial rule set based on the feature set, the initial rule set including multiple rules generated by decision paths; performing internal merging processing on the rules in the initial rule set to obtain a reference rule set, the internal merging processing including merging feature constraints with the same feature and retaining the intersection of constraint ranges; constructing a rule tree based on the reference rule set, recursively traversing each subtree in the rule tree, and extracting rules from each subtree that satisfy a first screening condition as target rules. Through structured merging and dynamic screening mechanisms, the method overcomes the path splitting limitations of traditional decision tree rule generation, achieving exponential compression of the number of rules while maintaining high accuracy and high recall.

The present application relates to the technical field of geotechnical reinforcement, in particular to a geocell mechanical property testing device of multi-cell interaction mechanism, comprising a pressure applying assembly, a sample assembly is arranged below the pressure applying end of the pressure applying assembly, the pressure applying assembly is used for applying pressure to the sample assembly, the sample assembly is electrically connected with a collection assembly, the collection assembly is used for real-time monitoring and recording parameters in the test process; the sample assembly comprises a model sample, and a bearing plate is arranged above and below the model sample; the model sample comprises a plurality of geocells, the plurality of geocells are stacked up and down, a strain gauge is arranged on each outer wall of the central cell of the geocell, and the strain gauges are electrically connected with the collection assembly. The present application can accurately simulate and measure the stress-strain characteristics under the multi-cell interaction mechanism, and provides important experimental basis and technical support for revealing the synergistic working mechanism of the reinforced structure and perfecting the geocell design theory.

Embodiments of the present application disclose a data processing method. The method comprises: acquiring a first image; obtaining a first correction matrix by means of a first model, wherein the first correction matrix is used for indicating noise of the first image; updating configuration information of a second model on the basis of the first correction matrix, wherein the configuration information is used for indicating a denoising strategy when the second model performs denoising processing; and performing denoising processing on the first image by means of the updated second model to obtain a second image. In the embodiments of the present application, the denoising strategy of the second model is adjusted by means of noise information of the input image, so that the second model samples noise more accurately during the denoising process, thereby improving the denoising capability of the model.

This application provides a meta-training method for a natural language query model, a natural language query method, a device, a storage medium, and a product. The meta-training method includes: classifying task types based on at least one of query type and query statement paradigm; collecting sample data from multiple task types to construct a first sample set and a second sample set for each task type; training a first model using a model-independent meta-learning algorithm based on the first sample set of multiple task types to obtain a second model; and fine-tuning the second model based on the second sample set of a single task type to obtain a third model for the corresponding task type. The sample data includes natural query statements annotated with structured query statements, and the first model is a pre-trained general natural language query model. This application combines transfer learning and meta-learning techniques and applies them to the field of natural language querying, effectively reducing model training costs while ensuring the accuracy of SQL statement conversion.

The present application relates to a kind of fractured bedrock buried hill oil reservoir gas injection parameter intelligent optimization method and system, belong to oil and gas field development field.The method includes establishing buried hill oil reservoir double medium model and buried hill oil reservoir embedded model;Multiple double medium model sample set and multiple embedded model sample set are obtained;The prediction set of double medium model and the prediction set of embedded model are obtained;Final prediction model is obtained;According to the prediction of each set of steam injection parameter in gas injection parameter combination according to final prediction model, obtain the predicted oil production set;Based on the predicted oil production set, obtain the optimized oil reservoir gas injection parameter.In the construction process of high-precision prediction model, the present application fuses the prediction set of double medium model and embedded fracture model as the sample set of prediction model, so that the result of prediction model is more in line with reality, solves the long time consumption of buried hill oil reservoir using different models to optimize gas injection parameter.

The application provides a GFS atmospheric precipitable water vapor enhancement method and device based on a BP neural network and a storage medium, and relates to the technical field of meteorological data processing. The method comprises the following steps: collecting multi-source basic data, extracting GFS original atmospheric precipitable water vapor PWV data, time and space geographic data, and combining to form a multi-dimensional input feature vector set; performing normalization processing on the feature vector set to obtain an input feature data set; performing time and space alignment processing on the GFS original PWV data and GNSS observation station data, taking the matched data as model samples, and dividing the model samples into a training set and a test set; constructing a BP neural network model, using an improved BP method with a momentum term to train and optimize an enhanced model; and inputting the test set into the model to output the enhanced atmospheric precipitable water vapor data after correction. The application realizes intelligent correction and precision improvement through a standardization process, does not require manual intervention, and can provide reliable support for meteorological monitoring, navigation positioning and climate analysis.

This application discloses a method, apparatus, device, and storage medium for resistivity and surface wave inversion probability fusion for shallow geological exploration, comprising: acquiring resistivity data and surface wave data of a target area; iteratively inverting and updating multiple model samples based on the resistivity data and surface wave data to determine the corresponding resistivity inversion model set information and velocity inversion model set information; evaluating the corresponding probability distribution and uncertainty voxels based on the resistivity inversion model set information and velocity inversion model set information to determine resistivity evaluation information and velocity evaluation information; determining the resistivity information entropy information and velocity information entropy information of uncertainty voxels at the corresponding normalized scale based on the resistivity information entropy information and velocity information entropy information; and fusing the resistivity evaluation information and velocity evaluation information based on the voxel weights corresponding to the resistivity information entropy information and velocity information entropy information to determine the fused probabilistic geological result.