A learning method for multi-label learning based on hashing method

Abstract

The invention discloses a multi-label learning design method based on a hashing method. Through the combination of a hashing algorithm and a multi-label learning algorithm based on Bayesian statistics, the correlation between labels is effectively utilized so as to improve the predicting performance of a multi-label learning model, labels and neighbors of the labels are introduced to computation of the posterior probability through the characteristics of the neighbors, the correlation between the labels is fully considered, and the accuracy of the algorithms is improved; the problem that the label space in multi-label learning of large-scale data is higher in dimension and sparse is solved through an MinHash algorithm; the purpose of learning large-scale data is achieved by finding the neighbors through locality sensitive hashing (LSH), the neighbors can be rapidly and efficiently found, and the expandability of the multi-label learning algorithm is improved.

Description

technical field [0001] The present invention relates to the technical field of machine learning, and in particular, to a learning method for multi-mark learning based on a hash method. Background technique [0002] In the traditional supervised learning framework, samples generally have a clear and single semantic label, that is, each sample example belongs to only one category. In this supervised learning framework, a variety of algorithms have been proposed and achieved good results. However, in many real-world applications, the semantic labels of research objects are usually not unique, and there are often cases where a sample can be assigned a set of multiple labels. For example, in text classification, a news report may cover multiple aspects of an event and, therefore, should be assigned to multiple topics (e.g., politics and economics); in bioinformatics, a gene or protein often has Multiple functions; in image annotation, an image can often be annotated with multipl...

AI Technical Summary

Problems solved by technology

[0011] The purpose of the present invention is to solve the problems encountered when the multi-label learning method is applied in a large-scale data scene, and propose a learning method based on hash method for multi-label learning. The method uses hash algorithm and Bayesian statistics Combined with the multi-label learning algorithm of learning, the correlation between labels is used to improve the prediction performance of the multi-label learning model; the MinHash algorithm is used to solve the problem that the label space is often more high-dimensional and sparse in the multi-label learning of large-scale data; Sensitive hashing (ie: LSH) performs neighbor search to solve the learning problem of large-scale data

Images