Constraining Variational Inference with Geometric Jensen-Shannon Divergence
We examine the problem of controlling divergences for latent space regularisation in variational autoencoders. Specifically, when aiming to reconstruct example x∈R^m via latent space z∈R^n (n≤ m), while balancing this against the need for generalisable latent representations. We present a regularisation mechanism based on the skew geometric-Jensen-Shannon divergence (JS^G_α). We find a variation in JS^G_α, motivated by limiting cases, which leads to an intuitive interpolation between forward and reverse KL in the space of both distributions and divergences. We motivate its potential benefits for VAEs through low-dimensional examples, before presenting quantitative and qualitative results. Our experiments demonstrate that skewing our variant of JS^G_α, in the context of JS^G_α-VAEs, leads to better reconstruction and generation when compared to several baseline VAEs. Our approach is entirely unsupervised and utilises only one hyperparameter which can be easily interpreted in latent space.
READ FULL TEXT