Generative Model without Prior Distribution Matching
This work addresses a specific bottleneck in generative modeling for researchers and practitioners using VAEs, offering an incremental improvement over existing methods.
The paper tackles the trade-off between reconstruction and generation in Variational Autoencoders (VAEs) by proposing a method where the prior matches the embedding distribution instead of forcing latent variables to fit a prior, resulting in improved preservation of the data manifold's geometric structure.
Variational Autoencoder (VAE) and its variations are classic generative models by learning a low-dimensional latent representation to satisfy some prior distribution (e.g., Gaussian distribution). Their advantages over GAN are that they can simultaneously generate high dimensional data and learn latent representations to reconstruct the inputs. However, it has been observed that a trade-off exists between reconstruction and generation since matching prior distribution may destroy the geometric structure of data manifold. To mitigate this problem, we propose to let the prior match the embedding distribution rather than imposing the latent variables to fit the prior. The embedding distribution is trained using a simple regularized autoencoder architecture which preserves the geometric structure to the maximum. Then an adversarial strategy is employed to achieve a latent mapping. We provide both theoretical and experimental support for the effectiveness of our method, which alleviates the contradiction between topological properties' preserving of data manifold and distribution matching in latent space.