spark.nn.neurons

Submodules

• spark.nn.neurons.adex
• spark.nn.neurons.alif
• spark.nn.neurons.base
• spark.nn.neurons.lif

Classes

Neuron	Abstract Neuron model.
NeuronConfig	Abstract Neuron model configuration class.
NeuronOutput	Generic Neuron model output spec.
ALIFNeuron	Leaky integrate and fire neuronal model.
ALIFNeuronConfig	ALIFNeuron configuration class.
AdExNeuron	Leaky Integrate and Fire neuronal model.
AdExNeuronConfig	AdExNeuron configuration class.

Package Contents

class spark.nn.neurons.Neuron(config=None, **kwargs)

Bases: spark.core.module.SparkModule, abc.ABC, Generic[ConfigT]

Abstract Neuron model.

This is a convenience class used to synchronize data more easily. Can be thought as the equivalent of Sequential in standard ML frameworks.

Parameters:

config (ConfigT | None)

config: ConfigT

units

reset()

Resets neuron states to their initial values.

abstractmethod __call__(in_spikes)

Execution method.

Parameters:

in_spikes (spark.core.payloads.SpikeArray)

Return type:

NeuronOutput

class spark.nn.neurons.NeuronConfig(__skip_validation__=False, **kwargs)

Bases: spark.core.config.SparkConfig

Abstract Neuron model configuration class.

Parameters:

__skip_validation__ (bool)

units: tuple[int, Ellipsis]

class spark.nn.neurons.NeuronOutput

Bases: TypedDict

Generic Neuron model output spec.

Initialize self. See help(type(self)) for accurate signature.

out_spikes: spark.core.payloads.SpikeArray

class spark.nn.neurons.ALIFNeuron(config=None, **kwargs)

Bases: spark.nn.neurons.Neuron

Leaky integrate and fire neuronal model.

Parameters:

config (ALIFNeuronConfig | None)

config: ALIFNeuronConfig

soma: spark.nn.components.somas.leaky.AdaptiveLeakySoma

delays: spark.nn.components.delays.base.Delays

synapses: spark.nn.components.synapses.base.Synanpses

learning_rule: spark.nn.components.learning_rules.base.LearningRule

build(input_specs)

Build method.

Parameters:

input_specs (dict[str, spark.core.specs.PortSpecs])

__call__(in_spikes)

Update neuron’s states and compute spikes.

Parameters:

in_spikes (spark.core.payloads.SpikeArray)

Return type:

spark.nn.neurons.NeuronOutput

class spark.nn.neurons.ALIFNeuronConfig(__skip_validation__=False, **kwargs)

Bases: spark.nn.neurons.NeuronConfig

ALIFNeuron configuration class.

Parameters:

__skip_validation__ (bool)

inhibitory_rate: float

soma: spark.nn.components.somas.leaky.AdaptiveLeakySomaConfig

synapses: spark.nn.components.synapses.base.SynanpsesConfig

delays: spark.nn.components.delays.base.DelaysConfig | None

learning_rule: spark.nn.components.learning_rules.base.LearningRuleConfig | None

class spark.nn.neurons.AdExNeuron(config=None, **kwargs)

Bases: spark.nn.neurons.Neuron

Leaky Integrate and Fire neuronal model.

Parameters:

config (AdExNeuronConfig | None)

config: AdExNeuronConfig

soma: spark.nn.components.somas.exponential.AdaptiveExponentialSoma

delays: spark.nn.components.delays.base.Delays

synapses: spark.nn.components.synapses.base.Synanpses

learning_rule: spark.nn.components.learning_rules.base.LearningRule

build(input_specs)

Build method.

Parameters:

input_specs (dict[str, spark.core.specs.PortSpecs])

__call__(in_spikes)

Update neuron’s states and compute spikes.

Parameters:

in_spikes (spark.core.payloads.SpikeArray)

Return type:

spark.nn.neurons.NeuronOutput

class spark.nn.neurons.AdExNeuronConfig(__skip_validation__=False, **kwargs)

Bases: spark.nn.neurons.NeuronConfig

AdExNeuron configuration class.

Parameters:

__skip_validation__ (bool)

inhibitory_rate: float

soma: spark.nn.components.somas.exponential.AdaptiveExponentialSomaConfig

synapses: spark.nn.components.synapses.base.SynanpsesConfig

delays: spark.nn.components.delays.base.DelaysConfig | None

learning_rule: spark.nn.components.learning_rules.base.LearningRuleConfig | None