This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
In this paper, we propose a new model that utilizes the potential of Convolutional Neural Network (CNN), considering the research trend that mainly relies on Transformer and MLP models in the existing long-term time series forecasting. We introduce a multi-scale time series reconstruction module that effectively captures the relationship between patches of different periods and the dependency between variables, and propose MS-TVNet, a multi-scale 3D dynamic CNN based on it. Experimental results on various datasets show that MS-TVNet outperforms existing models and achieves state-of-the-art (SOTA) results in long-term time series forecasting. This demonstrates the effectiveness of utilizing CNN to capture complex temporal patterns and suggests promising directions for future research in this field. The source code is available at https://github.com/Curyyfaust/TVNet .
