Neuromorphic Data Augmentation for Training Spiking Neural Networks
This addresses a critical issue for researchers developing neuromorphic intelligence with SNNs, offering a novel solution to improve training stability and generalization.
The paper tackles the problem of overfitting and unstable convergence in Spiking Neural Networks (SNNs) due to limited event-based datasets by proposing Neuromorphic Data Augmentation (NDA), resulting in accuracy gains of 10.1% on CIFAR10-DVS and 13.7% on N-Caltech 101.
Developing neuromorphic intelligence on event-based datasets with Spiking Neural Networks (SNNs) has recently attracted much research attention. However, the limited size of event-based datasets makes SNNs prone to overfitting and unstable convergence. This issue remains unexplored by previous academic works. In an effort to minimize this generalization gap, we propose Neuromorphic Data Augmentation (NDA), a family of geometric augmentations specifically designed for event-based datasets with the goal of significantly stabilizing the SNN training and reducing the generalization gap between training and test performance. The proposed method is simple and compatible with existing SNN training pipelines. Using the proposed augmentation, for the first time, we demonstrate the feasibility of unsupervised contrastive learning for SNNs. We conduct comprehensive experiments on prevailing neuromorphic vision benchmarks and show that NDA yields substantial improvements over previous state-of-the-art results. For example, the NDA-based SNN achieves accuracy gain on CIFAR10-DVS and N-Caltech 101 by 10.1% and 13.7%, respectively. Code is available on GitHub https://github.com/Intelligent-Computing-Lab-Yale/NDA_SNN