PredRNN: A Recurrent Neural Network for Spatiotemporal Predictive Learning

TL;DR

PredRNN introduces a novel recurrent neural network architecture with decoupled memory cells and bidirectional memory flow, significantly improving the modeling of complex spatiotemporal sequences for future frame prediction.

Contribution

It presents PredRNN with decoupled memory cells, zigzag memory flow, and a curriculum learning strategy, advancing the modeling of long-term spatiotemporal dependencies.

Findings

Achieved state-of-the-art results on five datasets.

Demonstrated effectiveness of memory decoupling and bidirectional flow.

Validated the approach's superiority in both action-free and action-conditioned scenarios.

Abstract

The predictive learning of spatiotemporal sequences aims to generate future images by learning from the historical context, where the visual dynamics are believed to have modular structures that can be learned with compositional subsystems. This paper models these structures by presenting PredRNN, a new recurrent network, in which a pair of memory cells are explicitly decoupled, operate in nearly independent transition manners, and finally form unified representations of the complex environment. Concretely, besides the original memory cell of LSTM, this network is featured by a zigzag memory flow that propagates in both bottom-up and top-down directions across all layers, enabling the learned visual dynamics at different levels of RNNs to communicate. It also leverages a memory decoupling loss to keep the memory cells from learning redundant features. We further propose a new curriculum learning strategy to force PredRNN to learn long-term dynamics from context frames, which can be generalized to most sequence-to-sequence models. We provide detailed ablation studies to verify the effectiveness of each component. Our approach is shown to obtain highly competitive results on five datasets for both action-free and action-conditioned predictive learning scenarios.