Minimal Convolutional RNNs Accelerate Spatiotemporal Learning

TL;DR

This paper introduces minimal convolutional RNN architectures that enable fully parallel training and achieve superior speed and accuracy in spatiotemporal forecasting tasks, combining efficiency with effective spatial modeling.

Contribution

The paper presents MinConvLSTM and MinConvGRU, novel convolutional RNNs that extend log-domain prefix-sum formulations for parallel training and incorporate exponential gating for improved efficiency.

Findings

Significantly faster training compared to standard ConvRNNs.

Lower prediction errors in Navier-Stokes and geopotential data.

Reduced parameter count and enhanced scalability.

Abstract

We introduce MinConvLSTM and MinConvGRU, two novel spatiotemporal models that combine the spatial inductive biases of convolutional recurrent networks with the training efficiency of minimal, parallelizable RNNs. Our approach extends the log-domain prefix-sum formulation of MinLSTM and MinGRU to convolutional architectures, enabling fully parallel training while retaining localized spatial modeling. This eliminates the need for sequential hidden state updates during teacher forcing - a major bottleneck in conventional ConvRNN models. In addition, we incorporate an exponential gating mechanism inspired by the xLSTM architecture into the MinConvLSTM, which further simplifies the log-domain computation. Our models are structurally minimal and computationally efficient, with reduced parameter count and improved scalability. We evaluate our models on two spatiotemporal forecasting tasks: Navier-Stokes dynamics and real-world geopotential data. In terms of training speed, our architectures significantly outperform standard ConvLSTMs and ConvGRUs. Moreover, our models also achieve lower prediction errors in both domains, even in closed-loop autoregressive mode. These findings demonstrate that minimal recurrent structures, when combined with convolutional input aggregation, offer a compelling and efficient alternative for spatiotemporal sequence modeling, bridging the gap between recurrent simplicity and spatial complexity.