The Thermodynamic Variational Objective
This addresses the challenge of improving variational inference for both continuous and discrete models in machine learning, offering a broadly applicable method with incremental advancements over existing techniques.
The paper tackles the problem of learning in deep generative models by introducing the Thermodynamic Variational Objective (TVO), which provides a tighter lower bound to the log marginal likelihood than the standard ELBO, and empirically demonstrates state-of-the-art performance in model and inference network learning.
We introduce the thermodynamic variational objective (TVO) for learning in both continuous and discrete deep generative models. The TVO arises from a key connection between variational inference and thermodynamic integration that results in a tighter lower bound to the log marginal likelihood than the standard variational variational evidence lower bound (ELBO) while remaining as broadly applicable. We provide a computationally efficient gradient estimator for the TVO that applies to continuous, discrete, and non-reparameterizable distributions and show that the objective functions used in variational inference, variational autoencoders, wake sleep, and inference compilation are all special cases of the TVO. We use the TVO to learn both discrete and continuous deep generative models and empirically demonstrate state of the art model and inference network learning.