Word embedding-based deep zero sample classification method

Abstract

The invention discloses a word embedding-based deep zero sample classification method. Types not appearing in a learning stage still can be identified in a prediction stage. A word embedding and depthfeature extraction-based zero sample image learning model is proposed; according to the method, a subspace of image feature and semantic feature joint embedding is learnt through a strong deep neuralnetwork; and by utilizing a semantic capability of word vectors, the purpose of zero sample learning of an image is achieved. A deep learning method and zero sample learning combination-based model is proposed; a large amount of improvements are made for a combination part embedding space; reference and guidance are provided for learning of image zero samples or zero samples in other modes; and the shortcomings of indefinite sample attribute definition, manual feature extraction and the like in conventional zero sample learning are overcome. The method is widely used in an end-to-end type prediction task, and is especially suitable for the type prediction task short of training samples and even short of a type of training samples.

Description

technical field [0001] The invention relates to the technical field of pattern recognition, in particular to a deep zero-sample classification method based on word embedding. Background technique [0002] The zero-shot learning (ZSL) model is a special machine learning problem, which belongs to the extreme case of transfer learning, that is, certain types of samples are missing in the training samples or certain types of samples never exist in the training data set. However, the actual situation requires us to It is still possible to identify class samples missing from these training sets in the test task. Zero-sample learning has a large number of applications in various real life. The most important application is to use it when there is not enough data. How to effectively improve the efficiency of zero-sample learning is of great significance. [0003] Due to the lack of samples in the training set, traditional machine learning methods cannot complete zero-shot learning ...

AI Technical Summary

Problems solved by technology

However, attribute-based zero-shot learning has essential shortcomings. The attributes of each category need to be set manually. Whether the classification effect is good or not depends on the quality of the artificial attribute selection. Different types of image classification require professionals in different fields. To do attribute calibration, and then to achieve classification, it is not an end-to-end implementation method

Moreover, artificially setting attributes will consume manpower and material resources, and the efficiency is not high

Images