A Variational Auto-Encoder Model for Stochastic Point Processes
This addresses the problem of probabilistic generative modeling for action sequences, which is incremental as it adapts VAE to point processes.
The paper tackles the challenge of modeling the variety of possible action sequences by proposing the Action Point Process VAE (APP-VAE), a variational auto-encoder that captures distributions over times and categories of actions, and validates it on MultiTHUMOS and Breakfast datasets.
We propose a novel probabilistic generative model for action sequences. The model is termed the Action Point Process VAE (APP-VAE), a variational auto-encoder that can capture the distribution over the times and categories of action sequences. Modeling the variety of possible action sequences is a challenge, which we show can be addressed via the APP-VAE's use of latent representations and non-linear functions to parameterize distributions over which event is likely to occur next in a sequence and at what time. We empirically validate the efficacy of APP-VAE for modeling action sequences on the MultiTHUMOS and Breakfast datasets.