-
Notifications
You must be signed in to change notification settings - Fork 3
Array conversions
This file has 3 methods for various conversions of the training data you record.
one hot means that only one value in the list can be 1 the params are:
filepath # the path which the new data will be saved
train_data # a npy array with the recorded data, the train data is a 2 by x array
# and on the first column are the images, on the second are the choices
# that were made in a form of a list.
This method just converts the choices in the given format:
A W S D A W S D WA WD SA SD
from choice = [x, x, x, x] to result_choice = [x, x, x, x, x, x, x, x]
and then creates a new npy array and saves it to the given path filepath.
the params are:
filepath # the path which the new data will be saved
train_data # a npy array with the recorded data, the train data is a 2 by x array
# and on the first column are the images, on the second are the choices
# that were made in a form of a list.
This method expect the choices to be in the form of long one hot array and will convert it to a shorter representation as follows:
This method just converts the choices in the given format:
A W S D WA WD SA SD A W D
from result_choice = [x, x, x, x, x, x, x, x] to choice = [x, x, x]
This method doesnt convert the S choice
the params are:
filepath # the path which the new data will be saved
train_data # a npy array with the recorded data, the train data is a 2 by x array
# and on the first column are the images, on the second are the choices
# that were made in a form of a list.
This methods expect the choices to be not one hot and will convert it to one hot array representation:
choice = [x, x, x] to one hot version result_choice = [x, x, x] # where only one values can
# be 1 and the others need to be zeroes
e.g(if i have pressed w and a the non one hot version will be choice = [1, 1, 0] and will be converted to [1, 0, 0]) The priority is at follows when converting A>D>W.