This paper addresses the challenge of blind non-linear equalization using a variational autoencoder (VAE) with a second-order Volterra channel model. The VAE framework’s costfunction, the evidence ...