A novel ergodic sequential logic spiking neural network with actor-critic reinforcement learning and its efficient FPGA implementation
Abstract
This paper presents a novel ergodic sequential logic (ESL) neuron model capable of mimicking typical excitabilities of biological neurons and realizing excitability suitable for artificial neural networks. Utilizing this model, a novel ESL spiking neural network is presented and its learning mechanism is also designed based on the actor-critic reinforcement learning framework. It is demonstrated that the presented network can enable an agent to solve a navigation task by providing appropriate actions. Additionally, comparisons indicate that the presented ESL spiking neural network is significantly more hardware-efficient than a traditional spiking neural network. Finally, potential applications for the ESL spiking neural network are discussed.
Fields of Study & Subjects
Advanced Memory and Neural Computing
Neural Networks and Applications
Neural dynamics and brain function
Publication details
- Journal / source
- IEICE Electronics Express
- Volume
- 23
- Issue
- 5
- Pages
- 20250727–20250727
- Publication date
- 18 January 2026
- Language
- EN