Poster
Filtering Variational Objectives
Chris Maddison · John Lawson · George Tucker · Nicolas Heess · Mohammad Norouzi · Andriy Mnih · Arnaud Doucet · Yee Teh

Tue Dec 5th 06:30 -- 10:30 PM @ Pacific Ballroom #114 #None
in Posters Tue »

When used as a surrogate objective for maximum likelihood estimation in latent variable models, the evidence lower bound (ELBO) produces state-of-the-art results. Inspired by this, we consider the extension of the ELBO to a family of lower bounds defined by a particle filter's estimator of the marginal likelihood, the filtering variational objectives (FIVOs). FIVOs take the same arguments as the ELBO, but can exploit a model's sequential structure to form tighter bounds. We present results that relate the tightness of FIVO's bound to the variance of the particle filter's estimator by considering the generic case of bounds defined as log-transformed likelihood estimators. Experimentally, we show that training with FIVO results in substantial improvements over training the same model architecture with the ELBO on sequential data.

Keywords: Generative Models · Variational Inference

Author Information

Chris Maddison (Oxford)
Dieterich Lawson (Google Brain)
George Tucker (Google Brain)
Nicolas Heess (Google DeepMind)
Mohammad Norouzi
Andriy Mnih (DeepMind)
Arnaud Doucet (Oxford)
Yee Teh (DeepMind)

More from the Same Authors