Cross-modal Hash model training method and device, cross-modal Hash model coding method and device and electronic equipment

Abstract

The invention provides a cross-modal Hash model training method, a cross-modal Hash model coding method, a cross-modal Hash model coding device and electronic equipment. The method relates to the technical field of artificial intelligence, and comprises the following steps: calling a cross-modal Hash model to carry out dimension reduction Hash coding processing on a plurality of obtained sample pairs to obtain a plurality of Hash coding pairs; for each Hash code pair, determining a target data Hash point with a relatively large weight in the Hash data point pair at each position in the Hash code pair, and determining a binary code of the Hash code pair based on each target Hash data point; based on each hash code pair, the similarity matrix corresponding to the plurality of sample pairs, and the difference between each hash code pair and the corresponding binary code, determining the total quantization loss of the cross-modal hash model; and updating parameters of the cross-modal hash model based on the total quantization loss. According to the invention, the coding precision of the cross-modal Hash model can be improved, and the computing resources occupied by the calculation of the similarity between the coding results of different samples can be saved.

Description

technical field [0001] The present application relates to artificial intelligence technology, and in particular to a training method, encoding method, device and electronic equipment of a cross-modal hash model. Background technique [0002] Artificial Intelligence (AI) is a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. [0003] Cross-modal hashing technology is a hash coding technology that maps high-dimensional data of different modalities to a unified low-dimensional Hamming space, usually through a cross-modal hashing model. When the cross-modal hash model converts high-dimensional data of samples of different modalities into low-dimensional hash coding, quantization loss will be generated. There is no more effective way to determine more accurate quantizati...

AI Technical Summary

Problems solved by technology

When the cross-modal hash model converts high-dimensional data of samples of different modalities into low-dimensional hash coding, quantization loss will be generated. There is no more effective way to determine more accurate quantization loss in related technologies, resulting in training-based The encoding accuracy of the cross-modal hash model is low

[0004] Since the cross-modal hash model processes high-dimensional data dimensionality reduction codes into hash codes with the same code length, and uses the distance between hash codes with the same code length to represent the similarity between samples (between hash codes The distance between the hash codes is negatively correlated with the similarity), and the accuracy of the distance between the hash codes to represent the similarity between the samples is not high when the coding accuracy is low. High computing resource consumption in the process of similarity between

Images