# A Variational Auto-Encoder Model for Stochastic Point Processes

## TL;DR

This paper introduces APP-VAE, a variational auto-encoder designed to model the complex distribution of action sequences over time and categories, effectively capturing their variability.

## Contribution

The paper presents a novel probabilistic generative model, APP-VAE, that uses latent representations and non-linear functions to model action sequence distributions.

## Key findings

- Effective modeling of action sequences demonstrated on MultiTHUMOS and Breakfast datasets.
- Captures the distribution over times and categories of actions.
- Addresses the challenge of modeling the variety of possible action sequences.

## Abstract

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.