PredRNN++: Towards A Resolution of the Deep-in-Time Dilemma in Spatiotemporal Predictive Learning

TL;DR

PredRNN++ introduces a novel recurrent network with causal LSTM units and a gradient highway architecture, significantly improving spatiotemporal modeling and long-term dependency capture in video prediction tasks.

Contribution

The paper proposes a new recurrent unit called causal LSTM and a gradient highway architecture to enhance deep-in-time video predictive learning.

Findings

Achieves state-of-the-art results on synthetic and real datasets.

Effectively alleviates the vanishing gradient problem.

Handles complex occlusion scenarios successfully.

Abstract

We present PredRNN++, an improved recurrent network for video predictive learning. In pursuit of a greater spatiotemporal modeling capability, our approach increases the transition depth between adjacent states by leveraging a novel recurrent unit, which is named Causal LSTM for re-organizing the spatial and temporal memories in a cascaded mechanism. However, there is still a dilemma in video predictive learning: increasingly deep-in-time models have been designed for capturing complex variations, while introducing more difficulties in the gradient back-propagation. To alleviate this undesirable effect, we propose a Gradient Highway architecture, which provides alternative shorter routes for gradient flows from outputs back to long-range inputs. This architecture works seamlessly with causal LSTMs, enabling PredRNN++ to capture short-term and long-term dependencies adaptively. We assess our model on both synthetic and real video datasets, showing its ability to ease the vanishing gradient problem and yield state-of-the-art prediction results even in a difficult objects occlusion scenario.