Tutorials

CompNeuro 101

Building spiking neural network models in GeNN

Neurons

Create a model consisting of a population of Izhikevich neurons with heterogeneous parameters, driven by a stimulus current. Simulate and record state variables.

Defining populations of neurons

Synapses

Create a simple balanced random network with two, sparsely connected populations of leaky integrate-and-fire neurons. Simulate and record spikes.

Adding synapses

MNIST inference

Perform MNIST inference by converting a pre-trained ANN to an SNN

Presenting a single image

Create a simple three layer network of integrate-and-fire neurons, densely connected with pre-trained weights. Present a single MNIST image and visualise spiking activity.

Classification of a single digit

Classifying entire test set

Present entire MNIST test set to previous model and calculate accuracy.

Classification of the entire test set

Improve classification performance

Use parallel batching and custom updates to improve inference performance by over 30x compared to previous tutorial.

Faster classification of the whole test set

Insect-inspired MNIST classification

Train a model of the insect mushroom body using an STDP learning rule to classify MNIST.

Projection Neurons

Create the first layer of Projection Neurons which convert input images into a sparse temporal code.

Presenting latency-coded inputs

Kenyon Cells

Add a second, randomly-connected layer of Kenyon Cells to the model.

Adding Kenyon Cells

Kenyon Cell gain control

Add recurrent inhibition circuit, inspired by <i>Giant GABAergic Neuron</i> in locusts, to improve sparse coding of the Kenyon Cells.

Feedback-inhibition based gain control

Mushroom Body Output Neurons

Add Mushroom Body Output Neurons with STDP learning and train model on MNIST training set.

Output neurons and learning

Testing

Create a simplified copy of the model without learning, load in the trained weights and calculate inference accuracy on MNIST test set.

Testing