Superposed Latent Autoencoder

Quanling Zhao, Jiaying Yang, Tianqi Zhang +4

Autoencoders typically meet tight latent-memory budgets by making each latent representation smaller, sacrificing representational capacity. We ask a different question: can multip…