Self-Organized Variational Autoencoders (Self-VAE) for Learned Image Compression
This work addresses image compression for applications requiring high efficiency, but it is incremental as it builds on existing variational autoencoder frameworks with a novel layer type.
The paper tackled the problem of improving learned image compression by proposing a Self-Organized Variational Autoencoder (Self-VAE) that replaces standard convolutional and GDN layers with self-organized operational layers, resulting in better rate-distortion performance and perceptual image quality.
In end-to-end optimized learned image compression, it is standard practice to use a convolutional variational autoencoder with generalized divisive normalization (GDN) to transform images into a latent space. Recently, Operational Neural Networks (ONNs) that learn the best non-linearity from a set of alternatives, and their self-organized variants, Self-ONNs, that approximate any non-linearity via Taylor series have been proposed to address the limitations of convolutional layers and a fixed nonlinear activation. In this paper, we propose to replace the convolutional and GDN layers in the variational autoencoder with self-organized operational layers, and propose a novel self-organized variational autoencoder (Self-VAE) architecture that benefits from stronger non-linearity. The experimental results demonstrate that the proposed Self-VAE yields improvements in both rate-distortion performance and perceptual image quality.