Rethinking the Role of Normalization and Residual Blocks for Spiking Neural Networks

• Main Authors: Ikegawa, S.-I (Author), Natori, N. (Author), Saiin, R. (Author), Sawada, Y. (Author)
• Format: Article
• Language: English
• Published: MDPI 2022
• Subjects: Biologically-inspired; Biomimetics; Chemical activation; Energy utilization; Energy-consumption; Neural networks; Neural-networks; Normalisation; normalization; Post-synaptic potentials; Power energy; Pre-activation residual block; pre-activation residual blocks; Simple++; Spiking neural network; spiking neural networks; Ultra-low power
• Online Access: View Fulltext in Publisher

 Biologically inspired spiking neural networks (SNNs) are widely used to realize ultralow-power energy consumption. However, deep SNNs are not easy to train due to the excessive firing of spiking neurons in the hidden layers. To tackle this problem, we propose a novel but simple normalization technique called postsynaptic potential normalization. This normalization removes the subtraction term from the standard normalization and uses the second raw moment instead of the variance as the division term. The spike firing can be controlled, enabling the training to proceed appropriately, by conducting this simple normalization to the postsynaptic potential. The experimental results show that SNNs with our normalization outperformed other models using other normalizations. Furthermore, through the pre-activation residual blocks, the proposed model can train with more than 100 layers without other special techniques dedicated to SNNs. © 2022 by the authors. Licensee MDPI, Basel, Switzerland.