Data compression method and device based on auto-encoder

Abstract

The invention discloses a data compression method and device based on an auto-encoder, and relates to the field of auto-encoders, the data compression device comprises an auto-encoder shell, the auto-encoder shell is of a cylindrical structure with an opening in one side, and an auto-encoder dismounting plate is fixed at the opening of the auto-encoder shell through a screw; the other side of the auto-encoder shell is integrally provided with an auto-encoder fixing plate, the auto-encoder fixing plate is provided with an auto-encoder fixing hole, and in order to solve the problems that when the auto-encoder is applied to an industrial motor, the overall equipment is likely to occupy a large area, the end of the auto-encoder shell is provided with a transmission connection assembly, and the transmission connection assembly is arranged on the end portion of the auto-encoder shell. The self-encoder fixing plate on the self-encoder shell can be fixed on one side, close to the motor rotating shaft, of the industrial motor through screws, so that the self-encoder shell is installed on the same side of the motor rotating shaft on the industrial motor, the occupied area is saved, the industrial motor shell does not need to be perforated, the service life of the industrial motor cannot be shortened, and the cost is reduced. And the use safety of the industrial motor is ensured.

Description

technical field [0001] The present invention relates to the field of self-encoders, in particular to a data compression method and device based on self-encoders. Background technique [0002] Autoencoders are a class of artificial neural networks used in semi-supervised learning and unsupervised learning, and their function is to learn representations of input information by taking the input information as a learning target. [0003] The autoencoder consists of an encoder and a decoder. According to the learning paradigm, autoencoders can be divided into systolic autoencoders, regular autoencoders and variational autoencoders, where the former two are discriminative models and the latter are generative models. Depending on the type of construction, the autoencoder can be a neural network with a feedforward structure or a recursive structure. [0004] Autoencoders function as representation learning algorithms in a general sense and are used for dimensionality reduction and...

AI Technical Summary

Problems solved by technology

A large amount of image data is generated and stored from the fields of daily life, social networking, public security monitoring, industrial production, etc., requiring a large amount of storage space

The currently popular mainstream lossy image compression format JPEG has poor compression rate and poor image quality. JPEG2000, which has not yet been popularized, may contain various patent disputes, and there is no obvious improvement compared to JPEG in various evaluation indicators. At the same time, the encoding and decoding speed Ten times slower and therefore not favored

Images