Omni-Scale CNNs: a simple and effective kernel size configuration for time series classification
This addresses a critical bottleneck in time series classification for researchers and practitioners by providing a universal method to capture optimal receptive field sizes across diverse datasets, though it is incremental as it builds on existing 1D-CNN frameworks.
The paper tackles the challenge of selecting appropriate receptive field sizes for 1D-CNNs in time series classification by proposing an Omni-Scale block with kernel sizes determined by a simple rule based on prime numbers, achieving state-of-the-art performance on four benchmarks.
The Receptive Field (RF) size has been one of the most important factors for One Dimensional Convolutional Neural Networks (1D-CNNs) on time series classification tasks. Large efforts have been taken to choose the appropriate size because it has a huge influence on the performance and differs significantly for each dataset. In this paper, we propose an Omni-Scale block (OS-block) for 1D-CNNs, where the kernel sizes are decided by a simple and universal rule. Particularly, it is a set of kernel sizes that can efficiently cover the best RF size across different datasets via consisting of multiple prime numbers according to the length of the time series. The experiment result shows that models with the OS-block can achieve a similar performance as models with the searched optimal RF size and due to the strong optimal RF size capture ability, simple 1D-CNN models with OS-block achieves the state-of-the-art performance on four time series benchmarks, including both univariate and multivariate data from multiple domains. Comprehensive analysis and discussions shed light on why the OS-block can capture optimal RF sizes across different datasets. Code available [https://github.com/Wensi-Tang/OS-CNN]