SSDP (Spike‑Synchrony‑Dependent Plasticity) is a biologically inspired learning rule that updates synaptic weights according to group‑level spike synchrony rather than isolated spike pairs.
This repository provides clean, task‑ready implementations of SSDP across three representative spiking‑neural‑network architectures:

Folder	Integration target	Brief description
SSD‑ViT/	SSD-ViT	Plug‑and‑play SSDP module added to the classifier and last‑stage DSSA projection weights
Single_layer_SSDP/	Custom single‑hidden‑layer SNN	Minimal example for quickly understanding and testing SSDP
SSDP_DH_SNN/	DH‑SNN (Dendritic & Heterogeneous SNN)	Demonstrates SSDP inside the publicly available DH‑SNN codebase
Visualization/	Analysis utilities	Scripts for t‑SNE, PCA, spike‑rate plots, etc.

Quick Start

2.1 SSD‑ViT

cd SSD-ViT python train.py --config cifar100.yaml

Dataset & config follow the original SpikingResformer repo. link: https://github.com/xyshi2000/SpikingResformer Only change: run train.py instead of main.py.

Where SSDP is inserted: classifier weights + last DSSA Wproj. Hyper‑parameters (A_plus, sigma, etc.) match those reported in the paper’s Supplementary Information.

2.2 Single‑layer SNN

cd Single_layer_SSDP python fashion_MNIST.py

2.3 SSDP with DH-SNN

Dataset & config follow the original DH-SNN repo. link: https://github.com/eva1801/DH-SNN

cd SSDP_DH_SNN python train_ssdp.py

License This project is released under the MIT License—see LICENSE for details.

Enjoy exploring synchrony‑driven learning! If you run into issues or have feature requests, feel free to email: [email protected].