Improvement of Spiking Neural Network with Bit Plane Coding

Spiking neural networks (SNNs) have gained sig-nificant attention as a promising framework in computational neuroscience and artificial intelligence, offering benefits such as low energy consumption and minimal memory requirements. Despite these advantages, their practical deployment is hindered by several challenges, particularly in performance optimization. This paper introduces a novel methodology to enhance the performance of SNNs in image processing by leveraging a bit-plane-based coding scheme. The proposed approach is specifically designed to improve the accuracy of SNNs without increasing their model size. Comprehensive experimental evaluations demonstrate the efficacy of this coding strategy across multiple datasets. To the best of our knowledge, this work is the first to explore the application of bit-plane representation in the context of SNNs.