File compression method based on inter-group loss optimization compression-friendly metric learning
The invention discloses a file compression method based on inter-group loss optimization compression-friendly metric learning. The method comprises the following steps: firstly, generating a file needing to be compressed into slices with a fixed size, constructing a file slice data set for training...
Saved in:
Main Authors | , , |
---|---|
Format | Patent |
Language | Chinese English |
Published |
05.01.2024
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | The invention discloses a file compression method based on inter-group loss optimization compression-friendly metric learning. The method comprises the following steps: firstly, generating a file needing to be compressed into slices with a fixed size, constructing a file slice data set for training and testing, obtaining implicit representations with the characteristics of'similar file data slices in the same group and different file slices in different groups' through deep metric learning, carrying out spectral clustering according to the implicit representations, merging the file slices in the same type, and carrying out spectral clustering; and high-compression-ratio compression of the file is realized. Meanwhile, due to the fact that the data structure needing to be stored and compressed is complex, the spectral clustering mode and the spectral clustering process do not have too many assumptions on the data structure, and a higher clustering speed can be achieved on a larger data set by constructing a spa |
---|---|
Bibliography: | Application Number: CN202311393490 |