Item Recommendation based on Multimedia Variational Autoencoder

At present, with the rapid development of science and technology and the rapid development of computer related technology, with the development of this series of technologies, multimedia communication has rapidly become the means of communication and communication. Digital image technology, a very important form of information representation in multimedia communication, has now become a mainstream way. The technology provides more convenience for people's lives, but also has many problems, such as security, transmission and storage. For example, some images will involve personal privacy, business secrets, and even government secrets. In addition, the rapid transmission of information in limited bandwidth is also very important. Especially in the presence of cloud storage, it can not only store large amounts of information, such as files, videos and images, but also provide a large enough online space to store shared data. In this case, the real-time and security requirements of information transmission are higher and stricter. It is of great theoretical and practical significance to compress the image and reduce the bandwidth and ensure the security of image encryption. Image encryption is also called digital image encryption. Image compression is mainly based on the use of different technologies to change the pixel value and location of the image to protect the image. Image compression mainly compresses the image by removing redundant or unrelated information. In this paper, the concept of the variational self encoder is introduced firstly, and then the image compression and encryption based on the variational self coder generation model are studied and summarized briefly.