Improving VAE generations of multimodal data through data-dependent conditional priors
This addresses a limitation in VAE-based generative modeling for multimodal data, but it is an incremental improvement focused on a specific bottleneck.
The paper tackled the problem of variational autoencoders (VAEs) being unable to generate samples from individual modalities in multimodal data due to the use of a simple isotropic Gaussian prior, and proposed CP-VAE, which learns data-dependent conditional priors to enable such generations, showing improved generative performance compared to baselines.
One of the major shortcomings of variational autoencoders is the inability to produce generations from the individual modalities of data originating from mixture distributions. This is primarily due to the use of a simple isotropic Gaussian as the prior for the latent code in the ancestral sampling procedure for the data generations. We propose a novel formulation of variational autoencoders, conditional prior VAE (CP-VAE), which learns to differentiate between the individual mixture components and therefore allows for generations from the distributional data clusters. We assume a two-level generative process with a continuous (Gaussian) latent variable sampled conditionally on a discrete (categorical) latent component. The new variational objective naturally couples the learning of the posterior and prior conditionals, and the learning of the latent categories encoding the multimodality of the original data in an unsupervised manner. The data-dependent conditional priors are then used to sample the continuous latent code when generating new samples from the individual mixture components corresponding to the multimodal structure of the original data. Our experimental results illustrate the generative performance of our new model comparing to multiple baselines.