Conditional Flow Variational Autoencoders for Structured Sequence Prediction

Apratim Bhattacharyya, Michael Hanselmann, Mario Fritz, Bernt Schiele, Christoph-Nikolas Straehle

TL;DR: We propose a conditional flow prior based CF-VAE for generative modelling of conditional distributions and effective regularisation schemes.

Abstract: Prediction of future states of the environment and interacting agents is a key competence required for autonomous agents to operate successfully in the real world. Prior work for structured sequence prediction based on latent variable models imposes a uni-modal standard Gaussian prior on the latent variables. This induces a strong model bias which makes it challenging to fully capture the multi-modality of the distribution of the future states. In this work, we introduce Conditional Flow Variational Autoencoders (CF-VAE) using our novel conditional normalizing flow based prior to capture complex multi-modal conditional distributions for effective structured sequence prediction. Moreover, we propose two novel regularization schemes which stabilizes training and deals with posterior collapse for stable training and better match to the data distribution. Our experiments on three multi-modal structured sequence prediction datasets -- MNIST Sequences, Stanford Drone and HighD -- show that the proposed method obtains state of art results across different evaluation metrics.

Code: https://drive.google.com/drive/folders/1L7RgmpA9j4gxF5hF8axRxDfQlI1xcfXo?usp=sharing

Keywords: Variational Inference, Normalizing Flows, Trajectories