encoding package

Submodules

cnsproject.encoding.encoders module

Module for encoding data into spike.

class cnsproject.encoding.encoders.AbstractEncoder(time: int, dt: Optional[float] = 1.0, device: Optional[str] = 'cpu', **kwargs)

Bases: abc.ABC

Abstract class to define encoding mechanism.

You will define the time duration into which you want to encode the data as time and define the time resolution as dt. All computations will be performed on the CPU by default. To handle computation on both GPU and CPU, make sure to set the device as defined in device attribute to all your tensors. You can add any other attributes to the child classes, if needed.

The computation procedure should be implemented in the __call__ method. Data will be passed to this method as a tensor for further computations. You might need to define more parameters for this method. The __call__ should return the tensor of spikes with the shape (time_steps, *population.shape).

Parameters

  • time (int) – Length of encoded tensor.

  • dt (float, Optional) – Simulation timestep. The default is 1.0.

  • device (str, Optional) – The device to do the comutations. The default is “cpu”.

class cnsproject.encoding.encoders.Time2FirstSpikeEncoder(time: int, dt: Optional[float] = 1.0, device: Optional[str] = 'cpu', **kwargs)

Bases: cnsproject.encoding.encoders.AbstractEncoder

Time-to-First-Spike coding.

Implement Time-to-First-Spike coding.

class cnsproject.encoding.encoders.PositionEncoder(time: int, dt: Optional[float] = 1.0, device: Optional[str] = 'cpu', **kwargs)

Bases: cnsproject.encoding.encoders.AbstractEncoder

Position coding.

Implement Position coding.

class cnsproject.encoding.encoders.PoissonEncoder(time: int, dt: Optional[float] = 1.0, device: Optional[str] = 'cpu', **kwargs)

Bases: cnsproject.encoding.encoders.AbstractEncoder

Poisson coding.

Implement Poisson coding.

Module contents