qqh-code/memristor-based-spiking-neural-network-accelerator-for-bio-inspired-interception-task

This repository contains the research code and documentation released alongside two academic papers on a specialized type of computer chip designed to mimic how biological brains process information. The chip is called a memristor-based spiking neural network accelerator, and the papers describe how it was designed and tested using a bio-inspired interception task: a scenario where a simulated predator-like agent tracks and responds to a moving target. That task was chosen because it is compact enough to analyze thoroughly while still requiring the kind of real-time, event-driven computation the hardware design was built around. The materials in this repository were produced during the research workflow rather than built as a general-purpose tool. They include MATLAB scripts for analyzing network behavior and simulation results, Verilog-A files that describe the electrical model of the memristor devices at the circuit level, sanitized templates for running circuit simulations using Cadence Spectre, a small set of publicly safe data examples, and figures from the papers. A significant portion of the original workflow depended on licensed Cadence tools and a 130 nm semiconductor process kit that are not included here and must be obtained separately. The Spectre templates use placeholder variables instead of the private paths from the research environment. The README is explicit that this is not a complete, runnable reproduction package: it is a curated public release meant to accompany citation of the papers rather than allow someone to replicate the full circuit design from scratch. The intended audience is researchers working at the intersection of neuromorphic hardware, spiking neural networks, and circuit simulation, who can supply their own licensed tool environment and use these files as a reference. The repository is released under the MIT license.