64 results about "Generalization error" patented technology

A boosting—based method and system for fusing a set of classifiers that performs classification using weak learners trained on different views of the training data. The final ensemble contains learners that are trained on examples sampled with a shared sampling distribution. The combination weights for the final weighting rule are obtained at each iteration based on the lowest training error among the views. Weights are updated in each iteration based on the lowest training error among all views at that iteration to form the shared sampling distribution used at the next iteration. In each iteration, a weak learner is selected from the pool of weak learners trained on disjoint views based on the lowest training error among all views, resulting in a lower training and generalization error bound of the final hypothesis.

The invention relates to a social media user demographic attribute prediction method based on multi-model stack fusion, and three demographic attributes including gender, age and region of a user arepredicted. The prediction of the three demographic attributes comprises the steps of (S1) user feature extraction, (S2) model training, and (S3) multi-model fusion to obtain a prediction result. The feature extraction in the present invention is not only for the text content of user Weibo, but also relates to statistical features, time information features and social relationship features, and theaccuracy of prediction is ensured. The mode of multi-model stack fusion is used to fuse logistic regression, random forest, and XGBoost models, a generalization error can be effectively reduced, andthe prediction accuracy is greatly improved.

A boosting—based method and system for fusing a set of classifiers that performs classification using weak learners trained on different views of the training data. The final ensemble contains learners that are trained on examples sampled with a shared sampling distribution. The combination weights for the final weighting rule are obtained at each iteration based on the lowest training error among the views. Weights are updated in each iteration based on the lowest training error among all views at that iteration to form the shared sampling distribution used at the next iteration. In each iteration, a weak learner is selected from the pool of weak learners trained on disjoint views based on the lowest training error among all views, resulting in a lower training and generalization error bound of the final hypothesis.

The invention discloses a method and device for predicting the operation stage and service life of a grounding grid of a substation based on the improved random forest algorithm. The method includes the following steps that initial data is obtained, and an original sample set is constructed; based on the characteristics of an original sample set, characteristic variables are extracted; the K-medoids method is used for clustering of the original sample set; the random forest algorithm is used for processing various samples, and a random forest model is formed; the to-be-predicted substation grounding grid characteristic variables are loaded into a random forest model, and the operation time in a feature vector is changed to obtain the relationship of an evaluation result and the operation time, and the predicted result of an operation stage and operation life are deduced. The randomness of the original sample set classification in the random forest algorithm is improved; the forest model is generated on the basis of the random forest algorithm, the generalization error is controllable, and the clustering accuracy is high; various factors affecting the state of the grounding grid ofthe substation are taken into account comprehensively, and different operation stages are divided to identify the corrosion state of the grounding grid in the corresponding stage through combination of the most suitable grounding grid fault detection method.

The invention discloses a training method and a detection method of a network traffic anomaly detection model. The network traffic anomaly detection model comprises a feature extraction network and aclassification network, and the training method comprises the following steps: determining the number of hidden layers and the number of neurons in each hidden layer according to a training sample; constructing an initial feature extraction network according to the number of the hidden layers and the number of neurons in each hidden layer; training the initial feature extraction network by using atraining sample to obtain a trained feature extraction network; extracting abstract feature data of a training sample by using the trained feature extraction network, and training a classification network by using the abstract feature data so as to complete training of a network traffic detection model. The network structure can adapt to network flow data, the situation that the structure of a detection model is too complex and too simple is avoided, and therefore, generalization errors are reduced, the detection time can be obviously shortened, and the detection accuracy can be obviously improved.

The invention relates to a wind speed ultra-short term prediction algorithm based on empirical mode decomposition and a random forest. The scale velocities at different time are obtained by decomposing historical real wind speed data to train a prediction model, during the wind speed prediction, according to present and previous moment wind speed data, the wind speed at the next moment can be predicated, and a predicated result of the wind speed at the next moment is obtained. The wind speed ultra-short term prediction algorithm has the advantages that the overfitting is not easy to happen, the generalization error is small, the algorithm is not sensitive to abnormal data, and the algorithm can be suitable for the practical production.

The invention discloses a power transformer fault diagnosis method. The power transformer fault diagnosis method comprises the following steps of determining N fault types of a transformer, and determining corresponding fault characteristic quantities for diagnosing the N fault types; taking the fault characteristic quantities corresponding to the N fault types as a testing sample, and performing normalizing processing of testing sample data; combining any two of the N fault types, establishing X (described in the specification) SVM ( Support Vector Machine) secondary classifiers, and training the X SVM secondary classifiers, and optimizing a SVM kernel function using a method based on combination of K-fold cross validation and an artificial bee colony algorithm; calculating generalization errors of each SVM classifier according to the K-fold cross validation method; and diagnosing the N fault types using an improved reordering adaptive directed acyclic graph support vector machine method. The invention has a capability of well diagnosing the fault type of a transformer, can greatly improve the accuracy of transformer fault diagnosis, and provides a reliable basis for transformer maintenance.

The invention provides a method for forecasting siRNA (Small interfering Ribonucleic Acid) interference efficiency during RNA (Ribonucleic Acid) interference, relates to the field of bioinformatics research, and aims to solve the problem of poor accuracy rate in existing methods for forecasting siRNA interference efficiency. The method comprises the following steps: inputting an siRNA sequence into a CPLD (Complex Programmable Logic Device); extracting a rule table according to features to re-encode the siRNA sequence, wherein as the siRNA interference efficiency is further influenced by a sequence and structural characteristics of a target mRNA (Messenger Ribonucleic Acid), the target mRNA is subjected to encoding while the siRNA sequence is encoded; and finally, using the random forest model to forecast the siRNA interference efficiency, during forecasting, using the known sample to establish an optimization model at first, carrying out OOB (out-of-bag) estimation by using OOB data generated by a bootstrap sampling method, enabling the generalization error of the model to be minimum by constantly mediating model parameters, and then using the optimized model to forecast the siRNA interference efficiency.