def test_if_values_is_valids():
v = np.array([
[0.2, 0.1, 0.1],
[0.3, 0.2, 0.1],
[0.3, 0.3, 0.2],
[0.2, 0.3, 0.3],
[0.1, 0.2, 0.3],
])
valueMax = 1
IC = layerF0(v, valueMax)
A = ARTFUZZY(IC)
A.train()
actual = [0.1, 0.1, 0.1, 0.7, 0.7, 0.7]
assert all([a == b for a, b in zip(actual, A.W[0])])
from src.neural_networks.art_fuzzy import ARTFUZZY
A = np.array([[0.25, 0.25], [0.25, 0.75], [0.75, 0.25], [0.75, 0.75]])
B = np.array([[0.25], [0.75], [0.75], [0.25]])
AC = layerF0(A, 1)
BC = layerF0(B, 1)
WAB = np.ones([AC.shape[0], BC.shape[0]])
rhoA = 0.5
rhoB = 0.9
rhoAB = 0.6
ArtA = ARTFUZZY(AC, rho=rhoA)
ArtB = ARTFUZZY(BC, rho=rhoB)
categoriesA = ArtA.categories()
categoriesB = ArtB.categories()
for i in range(0, len(ArtB.I)):
championB = max(categoriesB)
championIndexB = categoriesB.index(championB)
print()
if ArtB.hadRessonance(ArtB.I[i], ArtB.W[championIndexB]):
ArtB.W[championIndexB] = ArtB.learn(ArtB.I[i], ArtB.W[championIndexB])
def test_If_W_isintance_numpy():
A = ARTFUZZY([1.0, 2.0])
assert isinstance(A.I, np.ndarray)
from src.neural_networks.art_fuzzy import ARTFUZZY
from src.utils.functions import *
import numpy as np
v = np.array([
[0.2, 0.1, 0.1],
[0.3, 0.2, 0.1],
[0.3, 0.3, 0.2],
[0.2, 0.3, 0.3],
[0.1, 0.2, 0.3],
])
valueMax = 1
IC = layerF0(v, valueMax)
r = ARTFUZZY(IC)
r.train()
print(r.W)
print(r.Js)
x = []
for img_path in imgs_paths:
for i in imgs:
img = Image.open(img_path + str(i))
im = img.convert("L")
x.append(list(im.getdata()))
arr = np.array(x)
valueMax = 255
IC = layerF0(arr, valueMax)
print(IC)
r = ARTFUZZY(IC, rho=0.95)
r.train()
print()
#print(r.W)
print(r.Js)
circ_pesos = []
rect_pesos = []
for i in r.Js:
if i[0] < 40 and i[1] not in circ_pesos:
circ_pesos.append(i[1])
elif i[0] > 40 and i[1] not in rect_pesos:
rect_pesos.append(i[1])
class ARTMAPFUZZY(ART):
ArtA = []
ArtB = []
rho = 0
WAB = []
championsA = []
Map = []
def __init__(self, INPUT, OUTPUT, rhoINPUT=0.5, rhoOUTPUT=0.5, rho=0.5):
self.ArtA = ARTFUZZY(INPUT, rho=rhoINPUT)
self.ArtB = ARTFUZZY(OUTPUT, rho=rhoOUTPUT)
self.rho = rho
self.WAB = np.ones([INPUT.shape[0], OUTPUT.shape[0]])
def train(self):
print("Treinando ...")
for i in range(0, len(self.WAB)):
self.ArtB.match(i)
championIndexB = self.ArtB.getIndexOfChampion()
#print("-------", i)
#print(championIndexB, self.ArtB.getValueOfChampion())
categories = self.ArtA.categories()
championA = self.valueOfChampion(categories)
championIndexA = self.indexOfChampion(categories)
while championIndexA in self.championsA:
categories[championIndexA] = 0
championA = self.valueOfChampion(categories)
championIndexA = self.indexOfChampion(categories)
else:
self.championsA.append(championIndexA)
#print("Champion A:", championIndexA, championA)
while championA != 0:
#print("Vigilance test Art A: ", self.vigilanceValue(self.ArtA.I[i], self.ArtA.W[championIndexA]), self.rho)
if self.ArtA.hadRessonance(self.ArtA.I[i],
self.ArtA.W[championIndexA]):
#print("Vigilance test Match tracking: ", self.vigilanceValue(self.ArtB.Y[championIndexB], self.WAB[championIndexA]))
if self.hadRessonance(self.ArtB.Y[championIndexB],
self.WAB[championIndexA], self.rho):
self.ArtA.W[championIndexA] = self.ArtA.learn(
self.ArtA.I[i], self.ArtA.W[championIndexA])
self.ArtA.activate(championIndexA)
self.ArtA.Js.append([i, championIndexA])
#print("Ressonance!")
#print("Ativacao", self.ArtB.Y[championIndexB], self.WAB[championIndexA], championIndexB)
self.WAB[championIndexA] = self.activate(
self.WAB[championIndexA], championIndexB)
#print()
break
else:
categories[championIndexA] = 0
championA = self.valueOfChampion(categories)
championIndexA = self.indexOfChampion(categories)
x = self.AND(self.ArtA.I[i],
self.ArtA.W[championIndexA])
newRho = (sum(x) / sum(self.ArtA.I[i]))
#print(newRho, "NOVO", x, self.ArtA.I[i])
self.ArtA._rho = newRho
else:
#print("NAO PASSOU", championIndexA, championA)
categories[championIndexA] = 0
championA = self.valueOfChampion(categories)
championIndexA = self.indexOfChampion(categories)
def activate(self, W, i):
temp = np.zeros(len(W))
temp[i] = 1
return list(temp)
def test(self, INPUT, rho):
categories = self.ArtA.categories()
championA = self.valueOfChampion(categories)
championIndexA = self.indexOfChampion(categories)
while championA != 0:
if self.hadRessonance(INPUT, self.ArtA.I[championIndexA], rho):
return self.WAB[championIndexA]
else:
categories[championIndexA] = 0
championA = self.valueOfChampion(categories)
championIndexA = self.indexOfChampion(categories)
return -1
def testMapped(self, INPUT, rho):
categories = self.ArtA.categories()
championA = self.valueOfChampion(categories)
championIndexA = self.indexOfChampion(categories)
while championA != 0:
if self.hadRessonance(INPUT, self.ArtA.I[championIndexA], rho):
t = list(self.WAB[championIndexA])
artB = list(self.ArtB.W[t.index(1)])
s = [str(i) for i in artB]
return {
"index": t.index(1),
"value": self.WAB[championIndexA],
"ArtB": artB,
"id": "".join(s).replace(".", "")
}
else:
categories[championIndexA] = 0
championA = self.valueOfChampion(categories)
championIndexA = self.indexOfChampion(categories)
return -1
def __init__(self, INPUT, OUTPUT, rhoINPUT=0.5, rhoOUTPUT=0.5, rho=0.5):
self.ArtA = ARTFUZZY(INPUT, rho=rhoINPUT)
self.ArtB = ARTFUZZY(OUTPUT, rho=rhoOUTPUT)
self.rho = rho
self.WAB = np.ones([INPUT.shape[0], OUTPUT.shape[0]])
import numpy as np
from PIL import Image
from src.neural_networks.art_fuzzy import ARTFUZZY
img = Image.open(
r"C:/Users/ht3000052/Documents/artmap-fuzzy-tcc/samples/circ.png").convert(
'RGBA')
im = img.convert("L")
arr = np.array(list(im.getdata()))
arr = np.divide(arr, 255)
r = ARTFUZZY(arr)
print()
print(arr.size)
# arr = np.fromiter(iter(im.getdata()), np.uint8)
# arr.resize(im.height, im.width)
# arr ^= 0xFF # invert
# inverted_im = Image.fromarray(arr, mode='L')
# inverted_im.show()
quit()