evspikesim.SpikingNetwork

class evspikesim.SpikingNetwork

Spiking Neural Network (SNN) composed of layers of spiking neurons.

__init__(self: evspikesim.SpikingNetwork, compile_path: str = '/tmp/evspikesim') → None

Constructs an empty SNN.

Parameters

compile_path (str) – Compilation path for custom kernel callback sources.

Methods

add_fc_layer(self, n_inputs, n_neurons, ...)

Adds a fully-connected (FC) layer with the given arguments to the network.

add_fc_layer_from_source(self, n_inputs, ...)

Adds a layer of type LayerType with a custom kernel callbacks source.

infer(*args, **kwargs)

Overloaded function.

Attributes

output_layer

The output layer (i.e.
__getitem__(self: evspikesim.SpikingNetwork, index: int) → EvSpikeSim::Layer

Gets the layer at the specified index.

Parameters

index (int) – The index of the layer.

Returns

The requested layer.

Return type

Layer

__iter__(self: evspikesim.SpikingNetwork) → Iterator

Gets an iterator on the first layer.

Returns

An iterator on the first layer.

Return type

Iterator

add_fc_layer(self: evspikesim.SpikingNetwork, n_inputs: int, n_neurons: int, tau_s: float, threshold: float, initializer: EvSpikeSim::Initializer, buffer_size: int = 64) → EvSpikeSim::FCLayer

Adds a fully-connected (FC) layer with the given arguments to the network. The added layer uses the default kernel during inference.

Parameters

  • n_inputs (int) – The number of input neurons.

  • n_neurons (int) – The number of neurons in the layer.

  • tau_s (float) – The synaptic time constant of the neurons. The membrane time constant will be defined as twice tau_s.

  • threshold (float) – The threshold of the neurons.

  • initializer (Initializer.) – The initializer to use to initialize the weights of the layer.

  • buffer_size – The size (per neuron) of the post-synaptic spike times buffer used during inference.

Returns

The newly created FCLayer object.

Return type

FCLayer

add_fc_layer_from_source(self: evspikesim.SpikingNetwork, n_inputs: str, src_path: int, n_neurons: int, tau_s: float, threshold: float, initializer: EvSpikeSim::Initializer, buffer_size: int = 64) → EvSpikeSim::FCLayer

Adds a layer of type LayerType with a custom kernel callbacks source. If not already used, the given source file is compiled and loaded by the JITCompiler.

Parameters

  • src_path (str) – Path to the kernel callbacks source file.

  • n_inputs (int) – The number of input neurons.

  • n_neurons (int) – The number of neurons in the layer.

  • tau_s (float) – The synaptic time constant of the neurons. The membrane time constant will be defined as twice tau_s.

  • threshold (float) – The threshold of the neurons.

  • initializer (Initializer.) – The initializer to use to initialize the weights of the layer.

  • buffer_size – The size (per neuron) of the post-synaptic spike times buffer used during inference.

Returns

The newly created FCLayer object.

Return type

FCLayer

infer(*args, **kwargs)

Overloaded function.

  1. infer(self: evspikesim.SpikingNetwork, inputs: evspikesim.SpikeArray) -> evspikesim.SpikeArray

Infer the entire SNN using the given input spike array.

Parameters

inputs (SpikeArray.) – Sorted input spikes.

Returns

The output spike array.

Return type

SpikeArray

  1. infer(self: evspikesim.SpikingNetwork, indices: List[int], times: List[float]) -> evspikesim.SpikeArray

Infer the entire SNN using the given input spike indices and times.

Parameters

  • indices (List[int]) – Input spike indices.

  • times (List[float]) – Input spike times.

Returns

The output spike array.

Return type

SpikeArray

property output_layer

The output layer (i.e. the last layer of the network).