MesoNet: Diehl–Cook Brian2 baseline

keep the Diehl–Cook port nearby while changing the rule. trace input → conductance → spikes → updates, then compare

Chosen reference: Diehl–Cook 2015 Brian2 port. The journal's occasional “2005” wording conflicts with the repository/paper year; use the 2015 reference.

Why this route: the project needs explicit neuron and synaptic dynamics. The original code failed in the local Cython/compiler environment, while the updated port provides a more usable starting point. The notes retain alternatives such as symmetric STDP and BindsNET-based implementations.

Code navigation: Diehl&Cook_spiking_MNIST_B2.py for model/training logic; later weights/WeightReadout_B2.py for inspecting saved weights. These are reference-repository paths, not newly verified line locations in the local project.

XeAe input-to-excitatory weight matrix

Review questions: enumerate differences introduced by the port; fix the random seed and initialization policy; confirm that any reported baseline result uses unchanged encoding, inhibition and readout rules.

References saved in this journal section

• https://youtu.be/zldal7b7sJ4)
• https://snntorch.readthedocs.io/en/stable/tutorials/tutorial_5.html
• https://brian.discourse.group/t/unsupervised-learning-of-digit-recognition-using-spike-timing-dependent/189/23
• https://relatedrepos.com/gh/XiaoquinNUDT/Three-SNN-learning-algorithms-in-Brian2
• https://github.com/haoyz/sym-STDP-SNN