def __init__(self, input_count, rbf_count, output_count):
""" Create an RBF network with the specified shape.
@param input_count: The input count.
@param rbf_count: The RBF function count.
@param output_count: The output count.
"""
self.input_count = input_count
self.output_count = output_count
# calculate input and output weight counts
# add 1 to output to account for an extra bias node
input_weight_count = input_count * rbf_count
output_weight_count = (rbf_count + 1) * output_count
rbf_params = (input_count + 1) * rbf_count
self.long_term_memory = np.zeros(
(input_weight_count + output_weight_count + rbf_params),
dtype=float)
self.index_input_weights = 0
self.index_output_weights = input_weight_count + rbf_params
self.rbf = {}
# default the Rbf's to gaussian
for i in range(0, rbf_count):
rbf_index = input_weight_count + ((input_count + 1) * i)
self.rbf[i] = RbfGaussian(input_count, self.long_term_memory,
rbf_index)
def test_gaussian(self):
params = [5, 0, 0, 0]
funct = RbfGaussian(3, params, 0);
x = [-1, 0, 1]
y = funct.evaluate(x);
self.assertEquals(0.9607894391523232, y, 7)