32 results about "Nonlinear classifier" patented technology

A classifier training method and apparatus for training, a linear classifier trained by the method, and its use, are disclosed. In training the linear classifier, signatures for a set of training samples, such as images, in the form of multi-dimension vectors in a first multi-dimensional space, are converted to a second multi-dimension space, of the same or higher dimensionality than the first multi-dimension space, by applying a set of embedding functions, one for each dimension of the vector space. A linear classifier is trained in the second multi-dimension space. The linear classifier can approximate the accuracy of a non-linear classifier in the original space when predicting labels for new samples, but with lower computation cost in the learning phase.

The invention discloses a hydraulic cylinder fault diagnosis method. Firstly, the hydraulic cylinder is physically modeled and simulated in AMESim software and the fault injection process is completed to obtain fault data; then the wavelet packet energy is extracted from the data to construct a feature vector; and then the Fault diagnosis model based on high-dimensional nonlinear classifier; use training samples to train the fault diagnosis model, and use genetic algorithm to calculate the optimal solution of the parameters required by the model, and construct a multi-valued classifier; input hydraulic cylinder test samples to the model for further analysis Troubleshooting. The invention effectively solves the problem of insufficient fault data in the process of algorithm verification, and effectively solves the problem of nonlinearity and small samples by combining the application of a high-dimensional nonlinear classifier algorithm for optimized parameter search by genetic algorithm, and improves the fault identification ability of the fault diagnosis model.

Each embodiment of the invention generally relates to real-time rider detection by using synthetic training data, and relates to determining the existence of a rider at real-time in detail. The steps are as follows: receiving a target image; classifying the target image by using a linear classifier and determining an error value of the target image. If the error value is no more than a threshold, the classification is output. Or else, if the error value is larger than the threshold, the target is classified by using a nonlinear classifier.

The invention discloses a method and system for rapid electrocardiographic identification, wherein the method mainly includes a model training stage and a real-time testing stage. The present invention identifies and verifies personal identity information by collecting personal electrocardiographic signals, and belongs to In the field of biometric identification technology, first, any starting point position, any window larger than the length of a heartbeat, intercepts the ECG window, and divides each ECG window into fixed-length segments arbitrarily, and uses the fully automatic feature extraction layer to perform feature extraction; secondly , perform preliminary classification through a nonlinear classifier; then, generate a recognition result through the highest entropy voting; finally, realize fast ECG identity recognition through feature extraction and parallel processing of a nonlinear classifier, and the present invention does not require feature point detection and feature fusion , the identification speed is fast, and the ECG signal identification can be carried out in real time, and the verification crowd is wide.

This application provides a multi-task dictionary single classification method, including: obtaining tasks to be classified; learning a comprehensive dictionary and an analysis dictionary for each task to be classified; using the dictionary to learn models, analyze non-correlation items, and analyze coefficient coding The extraction item and the multi-task single classification item establish the objective optimization function; among them, the dictionary learning model includes a comprehensive dictionary and an analysis dictionary; the optimization function is solved to obtain a linear classifier and a nonlinear classifier respectively; the linear classifier and the nonlinear classifier are used to treat classification Tasks are categorized. This application uses a task to learn a comprehensive dictionary and an analysis dictionary, and makes the encoding coefficients as sparse as possible for other tasks, which can better represent the underlying structure of the data. At the same time, using the multi-task learning model also greatly reduces the computational complexity. The present application also provides a multi-task dictionary single classification system, a device and a computer-readable storage medium, which have the above-mentioned advantageous effects.