Abstract:Spiking Neural Networks (SNNs) are capable of encoding and processing temporal information in a biologically plausible way. However, most existing SNN-based methods for image tasks do not fully exploit this feature. Moreover, they often overlook the role of adaptive threshold in spiking neurons, which can enhance their dynamic behavior and learning ability. To address these issues, we propose a novel method for image decoding based on temporal attention (TAID) and an adaptive Leaky-Integrate-and-Fire (ALIF) neuron model. Our method leverages the temporal information of SNN outputs to generate high-quality images that surpass the state-of-the-art (SOTA) in terms of Inception score, Fréchet Inception Distance, and Fréchet Autoencoder Distance. Furthermore, our ALIF neuron model achieves remarkable classification accuracy on MNIST (99.78\%) and CIFAR-10 (93.89\%) datasets, demonstrating the effectiveness of learning adaptive thresholds for spiking neurons. The code is available at <a class="link-external link-https" href="https://github.com/bollossom/ICLR_TINY_SNN" rel="external noopener nofollow">this https URL</a>.

What problem does this paper attempt to address?

The main problem this paper attempts to address is that existing methods based on Spiking Neural Networks (SNNs) fail to fully utilize temporal information in image tasks and ignore the enhancement effect of adaptive thresholds on the dynamic behavior and learning ability of spiking neurons. Specifically: 1. **Insufficient utilization of temporal information**: Most existing SNN methods do not fully exploit and utilize the rich temporal information in the SNN output when processing image tasks. 2. **Neglect of adaptive thresholds**: Existing SNN methods usually ignore the impact of adaptive thresholds on spiking neurons, while adaptive thresholds can significantly improve the dynamic behavior and learning ability of neurons. To address these issues, the authors propose two new methods: - **Temporal Attention Image Decoding (TAID)**: By introducing a temporal attention mechanism, the temporal relationships between different output spike sequences are established, thereby generating high-quality images. - **Adaptive Leaky-Integrate-and-Fire (ALIF) neuron model**: By making the time constant \(\tau\) and threshold \(V_{th}\) learnable, the dynamic behavior and learning ability of SNNs are enhanced. Experimental results on multiple datasets show that these methods achieve significant performance improvements in both image generation and classification tasks. For example, on the MNIST and CIFAR-10 datasets, the ALIF model achieved classification accuracies of 99.78% and 93.89%, respectively, while in image generation tasks, the TAID method outperformed existing state-of-the-art methods on multiple evaluation metrics.

When Spiking neural networks meet temporal attention image decoding and adaptive spiking neuron