33 results about "Correlation clustering" patented technology

A framework for an automated news recommendation system for financial analysis. The system includes the automated ingestion, relevancy, clustering, and ranking of news events for financial analysts in the capital markets. The framework is adaptable to any form of input news data and can seamlessly integrate with other data used for analysis like financial data.

Disclosed is a rolling bearing remaining life prediction method based on feature fusion and particle filtering. According to an index calculation process, firstly, original features are extracted from bearing vibration signals, the extracted original features are clustered by the adoption of a relevance clustering method, then, one typical feature is selected from each cluster to form optimal feature sets, and finally the feature sets are fused by the adoption of a weight fusion method into a final recession index. According to a life prediction process, firstly, smoothing and resampling are carried out on the recession index, the time interval is adjusted to be an expected value, state-space model initial parameters are calculated by the adoption of least square fitting, then, model parameters are updated in real time according to new observation data, and finally the remaining life of a bearing can be predicted. According to the rolling bearing remaining life prediction method based on feature fusion and particle filtering, the difference between the life prediction result and a true value is small, and the application effect is good.

The invention discloses an iterative hyperspectral image lossless compression method based on low-rank representation, which solves the problems that the traditional compression method ignores the correlation of image space, the clustering result is unstable, and there is no connection between modules. The implementation steps include: defining the spectral angle similarity measurement method; roughly clustering the original image; solving the rough clustering block coefficient matrix by low-rank representation; re-clustering the coefficient matrix to obtain the initial clustering result; iteratively optimizing the initial clustering result to obtain The prediction coefficient and prediction residual of the final clustering block; entropy coding is then performed to obtain the code stream file to be transmitted; after entropy decoding, the code stream file is decompressed at the decoding end to obtain a lossless compressed hyperspectral image. The invention defines a spectral angle correlation measurement method to increase the utilization of spatial correlation; the combination of low-rank representation and subspace clustering increases the stability of clustering results; and iteratively optimizes and correlates each module to increase the result compression ratio. Used in the field of image compression.

The invention relates to a method, system, device and medium for forecasting wireless traffic on holidays based on a correlation clustering hybrid algorithm model. The forecasting method for wireless traffic on holidays includes: step 1) defining holiday weights according to different dates; step 2) according to holidays Weight calculation of the holiday coefficient of the base station sector; step 3) classifying the traffic patterns of all base station sectors according to the holiday coefficient; step 4) constructing a hybrid algorithm model; step 5) obtaining the well-trained hybrid algorithm model corresponding to different traffic patterns; Step 6) Use the trained hybrid algorithm model to predict the wireless traffic of the base station sector. The present invention proposes to divide the base station sectors into different traffic patterns by holiday coefficients, and train a common model framework for each pattern, which can effectively solve the problem of over-fitting, and make the model have flexibility and the ability to capture the characteristics of traffic changes.

The invention provides enterprise hidden danger management, including an enterprise safety detection standard library, a post customization detection system, and a hidden danger diagnosis and analysis system; the invention also relates to an enterprise hidden danger management method, electronic equipment and a storage medium. The present invention uses wireless positioning and mobile applications to complete the point-to-point positioning of inspection personnel, automatic triggering of inspection content, and customized push of inspection messages. The present invention realizes the real-time collection of full-staff information, processes the unstructured data of safety production through semantic analysis technology, classifies and summarizes each element of safety production big data, uses association rules to analyze the correlation between each element, and clusters to find out potential dangers Factors, establish a safety production risk analysis model suitable for hidden danger investigation business data, control the unsafe behavior of people, the unsafe state of objects, the unsafe state of machines, the unsafe factors of the environment, and the lack of management. The relationship between the five elements of environmental management to achieve the purpose of preventing and reducing safety production accidents.

The invention discloses a traffic flow forecasting method based on road clustering and double-layer bidirectional LSTM, including: 1) When there is a missing value in the training data, the missing value is filled in the way of surrounding average for the missing value, Improve prediction accuracy; 2) Propose a method of correlation clustering of roads based on historical traffic data, divide roads into several groups, and use time information and spatial information at the same time in the data preprocessing stage to improve prediction accuracy; 3) Design a A two-layer bidirectional LSTM deep neural network model, which improves the prediction accuracy of the model; 4) proposes a method for batch training and testing the network model to speed up the training and testing speed of the neural network model; 5) proposes a Multi-model fusion method to improve prediction accuracy. The invention simultaneously improves the prediction speed and accuracy of the deep neural network in traffic flow prediction.

The invention belongs to the technical field of information, and particularly relates to a user cost configuration method and a user cost configuration system. The user cost configuration method comprises the steps of: S1, acquiring network data of users, and carrying out service analysis on the network data of the users; S2, generating service vector sets corresponding to the users according to service types of the network data of the users, carrying out correlation clustering on users with the same internet surfing preference, and carrying out preference classification on the service types of the users; S3, according to service preference types of the users, calculating convergence groups of different service preference types in a set time period and service characteristics correspondingto the convergence groups; and S4, according to the service characteristics corresponding to the convergence groups, configuring package modes to the users. According to the user cost configuration method and the user cost configuration system, a charge mode of dynamically sensing the service preferences of the users is implemented, and experience of the users can be greatly improved.