Classification model training method and device

Abstract

The invention provides a classification model training method and device, and the method comprises the steps: calculating the prior probability of each sample feature in N different classification types, wherein N is greater than 1; for each sample feature, determining a classification coefficient of the sample feature according to a prior probability of the sample feature in N different classification categories; forming a classification coefficient matrix by the classification coefficients of the sample features, and determining features of a sample to be trained according to the classification coefficient matrix and a sample feature matrix, the sample feature matrix being composed of the sample features; and training the features of the to-be-trained sample by adopting a GBDT algorithmto obtain a classification model. According to the method and device, the conversion of the sample features can be realized, new features are provided for iteration during training of the classification model, and the construction of feature engineering is facilitated, so that the classification effect of the classification model is improved.

Description

technical field [0001] The present application relates to the technical field of machine learning, in particular to a method and device for training a classification model. Background technique [0002] Supervised learning is a method in machine learning, which can learn or establish a learning model from training data, and infer new instances based on this model. Among them, supervised learning methods can be divided into generative methods (also known as generative learning algorithms, generative learning algorithm) and discriminative methods (also known as discriminative learning methods, discriminative learning algorithm), and the learned models are called generative models (Generative Model ) and discriminative model (Discriminative Model). [0003] In the generation method, the joint probability density distribution P(X,Y) is learned from the data, and then the conditional probability distribution P(Y|X) is obtained as the predicted model, that is, the generation mode...

AI Technical Summary

Problems solved by technology

In this case, the discriminative method cannot be used

[0007] The characteristics of the discriminant method: the discriminant method directly learns the decision function Y=f(X) or the conditional probability distribution P(Y|X), which cannot reflect the characteristics of the training data itself

Images