neuron_traits library

The neuron traits library is a set of neuron attributes. You can use the neuron attributes provided in the library, for example, when constructing populations.

Currently, the library supports the property sets for the following neurons:

• Neuron based on the AltAILIF model

  The neuron based on the AltAILIF model is a leaky integrator neuron (also known as a leaky integrate-and-fire neuron). The neuron is characterized by the membrane potential and the threshold values of the membrane potential.

  The value of the membrane potential changes in accordance with the value of the synaptic impact coming from the associated synapse. If the membrane potential of a neuron reaches a positive or negative threshold value, the neuron generates a positive or negative spike, respectively. Depending on the specified attributes of the neuron, after one of the thresholds is exceeded, the membrane potential either takes on the value by which the threshold was exceeded, or is reset to a specified value.

  

  Neuron junction with the main function to transmit a signal from one neuron to another. Each neuron can be connected to multiple synapses.

  

  A value generated by a synapse for changing the values of the neuron attributes.

  The value of the membrane potential also changes linearly by the leakage value potential_leak_, which can be either positive or negative.

  In the attributes of a neuron instance, you can specify the baseline value of the membrane potential, the value of the membrane potential after the reset, a positive and negative threshold value of the potential, and the leakage value; you can also set flags that control the behavior of the potential after the thresholds are exceeded, a flag for resetting the neuron potential at the beginning of each time step, and a flag controlling the leakage behavior depending on the current neuron potential sign.

• Neuron based on the BLIFATNeuron model ("BLIFAT neuron").

  A neuron based on the BLIFATNeuron model is an integrator neuron with a leak and an adaptive threshold, which generates a series of spikes (Bursting Leaky Integrate-and-Fire with Adaptive Threshold Neuron). A BLIFAT neuron is characterized by the membrane potential and the threshold value of the membrane potential.

  

  Short-term action potential generated by a neuron in a population as a result of the neuron attribute changes.

  The value of the membrane potential changes in accordance with the value of the synaptic impact coming from the associated synapse. If the neuron membrane potential reaches a threshold value, the neuron starts generating a series of spikes with a certain interval. In this case, after the spike generation, the membrane potential tends to the base value, and the threshold value changes in accordance with the increment.

  If a neuron does not receive messages from a connected synapse for a certain time, the membrane potential and its threshold value exponentially tend to the base values.

  You can specify the base value of the membrane potential, the base threshold value, the increment value, the time interval upon reaching which the values of the membrane potential and potential threshold begin to approach the base value, the duration of a spike series, and the interval for spike generation in the attributes of a neuron instance.

• Neuron based on the SynapticResourceSTDPBLIFATNeuron model.

  A neuron based on the SynapticResource STDPBLIFATNeuron model is a BLIFAT neuron characterized by a free synaptic resource, a threshold value of a free synaptic resource, and stability.

  If the absolute value of the free synaptic resource exceeds the threshold value, this free synaptic resource is distributed equally among all plastic presynaptic connections of the neuron. After that, the value of the free synaptic resource of the neuron equals 0. If the absolute value of the free synaptic resource is less than the threshold value, the neuron appropriates the synaptic resource from connected synapses until it reaches its threshold value.

  The neuron stability value changes in accordance with the synaptic plasticity used by the synapse and affects its plasticity value.

For more details on the properties of neurons, see API documentation for Kaspersky Neuromorphic Platform. If necessary, you can implement and use your own neuron models when working with the source code.

Page top