About Kaspersky Neuromorphic Platform backend

The platform provides several independent backends for execution of spiking neural networks on various computing devices.

The Kaspersky Neuromorphic Platform backends perform the following functions:

Start spiking neural networks on various computers.
A device or code that directly executes a network.
Perform static verifications of the spiking neural network design.
Data exchange between the neural network and the external environment.
A set of elements external to the Kaspersky Neuromorphic Platform with which the platform interacts.

You can use the following platform components to implement backends:

core component. Provides classes, such as projection and population classes, required for interacting with a backend. The component is supported for C++ and Python.
devices component. Provides a set of interfaces for interacting with physical devices. The component is supported for C++.
neuron_traits component. Contains neuron properties. The component is supported for C++ and Python.
A neural network node with a set of attributes. Based on the neuron attributes and the sequence of input signals, a certain mathematical function is calculated. The result of the function is a synaptic impact or its absence.
synapse_traits component. Contains synapse model properties. The component is supported for C++ and Python.
Neuron junction with the main function to transmit a signal from one neuron to another. Each neuron can be connected to multiple synapses.
meta component. Provides general-purpose templates. The component is supported for C++.

Do not use a framework component to implement backends to avoid circular dependencies between the framework and backends.

In this section