TVNet: A Novel Time Series Analysis Method Based on Dynamic Convolution and 3D-Variation

At present, the research on time series often focuses on the use of Transformerbased and MLP-based models.Conversely, the performance of Convolutional Neural Networks (CNNs) in time series analysis has fallen short of expectations, diminishing their potential for future applications. Our research aims to enhance the representational capacity of Convolutional Neural Networks (CNNs) in time series analysis by introducing novel perspectives and design innovations. To be specific, We introduce a novel time series reshaping technique that considers the inter-patch, intra-patch, and cross-variable dimensions. Consequently, we propose TVNet, a dynamic convolutional network leveraging a 3D perspective to employ time series analysis. TVNet retains the computational efficiency of CNNs and achieves state-of-the-art results in five key time series analysis tasks, offering a superior balance of efficiency and performance over the state-of-theart Transformer-based and MLP-based models. Additionally, our findings suggest that TVNet exhibits enhanced transferability and robustness. Therefore, it provides a new perspective for applying CNN in advanced time series analysis tasks. * Equal contribution. † Corresponding author. '/LQHDU *%V L7UDQVIRUPHU *%V 791HW2XUV *%V &URVVIRUPHU *%V )('IRUPHU *%V 7LPHV1HW *%V 7LPH0L[HU *%V (77P9DULDEOHV/ Memory Footprint (GB) 2.2GB 5.2GB 7.2GB 50 100 150 200 250 300 350 400 7UDLQLQJ7LPHVHSRFK 0.38 0.39 0.40 0.41 0.42 0$( 3DWFK767 *%V 5/LQHDU *%V 0RGHUQ7&1 *%V 791HW2XUV *%V 0,&1 *%V )('IRUPHU *%V 7LPHV1HW *%V 7LPH0L[HU *%V