Discrete memristive spiking neural networks: investigating information flow, synchronization, and emergent...

The processing of information within complex neural networks is a challenge topic that has intrigued researchers for many years. In this paper, we conducted an in-depth investigation into the learning mechanisms that are intrinsic to discrete memristor spiking neural networks. We also explored the effectiveness of information transmission and synchronization among various neurons and networks. Firstly, a memristor model with memory regulation function and tanh function's nonlinear characteristics was constructed. This model not only ensures that the internal state variables of the memristor do not exhibit divergence, but also demonstrates that this memristor is suitable for spiking signal processing and has the ability to transmit spiking signals. Secondly, our research delved into the intricate dynamics of these discrete spiking neural networks, including the ternary coupled spiking neural network and ring coupled spiking neural network, aiming to shed light on how they operate and interact. Thirdly, based on the designed pulse neurons, this study constructed a simple pulse neuron network. By reasonably setting the relevant parameters, the research found that this network possesses the ability for pattern recognition. The results of our investigation are crucial for understanding the mechanisms of information processing and synchronization phenomena within neural networks. It provides valuable insights into the potential of memristor networks in advancing artificial intelligence and computational neuroscience.