Optimized Spiking Neural Network Architecture for Fashion MNIST Classification: A Comparative Study with Convolutional Neural Networks

The Spiking Neural Network (SNN) is an event-driven and energy-efficient system inspired by biological neurons. It began to appeal too much in the past few years. The optimized SNN architecture holds a new record for Fashion MNIST classification and has structural innovations to enhance performance. Comparison with Convolutional Neural Networks (CNNs) Costs of operation and image processing points to a result in favour of SNNs. Experimental results show that the optimized SNN obtains competitive classification accuracy with significantly lower energy consumption than previous efforts, making it ideal for anything from real-time to yearlong power efficiency applications.