We study the effects of quantization and additive white Gaussian noise (AWGN) in transmitting latent representations of im-ages over a noisy communication channel. The latent representations are obtained using autoencoders (AEs). We analyze image recon-struction and classification performance for different channel noise powers, latent vector sizes, and number of quantization bits used for the latent variables as well as AEs' parameters. The results show that the digital transmission of latent representations using conventional AEs alone is extremely vulnerable to channel noise and quantization effects. We then pro-pose a combination of basic AE and a denois-ing autoencoder (DAE) to denoise the corrupt-ed latent vectors at the receiver. This approach demonstrates robustness against channel noise and quantization effects and enables a signif-icant improvement in image reconstruction and classification performance particularly in adverse scenarios with high noise powers and significant quantization effects.
展开▼
