wisardpkg

KernelCanvas

constructor:

numberOfKernels = 10
dimension = 2

kc = KernelCanvas(
    dimension,               # required
    numberOfKernels,         # required
    bitsBykernel = 3,        # optional
    activationDegree = 0.07, # optional
    useDirection = False     # optional
)

• dimension: [int] this is the dimension of your input.
• numberOfKernels: [int] this the number of kernel you wish to create.
• bitsByKernel: [int] this is the number of bits by kernel in the output data.
• activationDegree: [float] this is the proportion of how many kernels will be activated.
• useDirection: [boolean] this activate the use of direction of the data.

methods:

transform

This method receive a list of points in double precision with dimension passed in the creation of KernelCanvas object.

from random import random

kc = KernelCanvas(10,2,3)

sequenceData = []
for i in range(100):
    point = [0.0,0.0]
    point[0] = 10*random()
    point[1] = 10*random()
    sequenceData.append(point)

out = kc.transform(sequenceData)

It will return the list of new input mapped from kernel. It will just zeros and ones of size numberOfKernels * bitsBykernel

This site is open source. Improve this page.