Convolutional Tensor-Train LSTM for Spatio-temporal Learning
This addresses the problem of poor performance in long-term video forecasting for applications like human-behavior analysis and object tracking, representing an incremental improvement through a novel decomposition method.
The paper tackles the challenge of learning long-term spatio-temporal correlations in video sequences for tasks like forecasting, proposing a higher-order convolutional LSTM with a tensor-train module to efficiently capture these correlations using fewer parameters. It achieves state-of-the-art performance on datasets such as Moving-MNIST-2, KTH action, and Something-Something V2.
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.