Convolutional Tensor-Train LSTM for Spatio-temporal Learning

TL;DR

This paper introduces a convolutional tensor-train LSTM model that efficiently captures long-term spatio-temporal correlations in video data, achieving state-of-the-art results with fewer parameters.

Contribution

It proposes a novel tensor train decomposition for convolutional LSTMs, enabling efficient learning of long-term dependencies in video sequences.

Findings

Outperforms existing methods on video prediction tasks

Uses significantly fewer parameters than baseline models

Achieves state-of-the-art results on multiple datasets

Abstract

Learning from spatio-temporal data has numerous applications such as human-behavior analysis, object tracking, video compression, and physics simulation.However, existing methods still perform poorly on challenging video tasks such as long-term forecasting. This is because these kinds of challenging tasks require learning long-term spatio-temporal correlations in the video sequence. In this paper, we propose a higher-order convolutional LSTM model that can efficiently learn these correlations, along with a succinct representations of the history. This is accomplished through a novel tensor train module that performs prediction by combining convolutional features across time. To make this feasible in terms of computation and memory requirements, we propose a novel convolutional tensor-train decomposition of the higher-order model. This decomposition reduces the model complexity by jointly approximating a sequence of convolutional kernels asa low-rank tensor-train factorization. As a result, our model outperforms existing approaches, but uses only a fraction of parameters, including the baseline models.Our results achieve state-of-the-art performance in a wide range of applications and datasets, including the multi-steps video prediction on the Moving-MNIST-2and KTH action datasets as well as early activity recognition on the Something-Something V2 dataset.