def transpose(M):
if M == []:
cm('Matrix M is empty', 'fail')
return []
rows = len(M)
try:
cols = len(M[0])
except TypeError:
cols = 1
# Transpose M
T = list()
for col in range(cols):
A = list()
for row in range(rows):
if rows == 1 and cols == 1:
M = clean(M)
A.append(M[0])
elif rows != 1 and cols == 1:
A.append(M[row])
else:
A.append(M[row][col])
if len(A) == 1:
T.append(A[0])
else:
T.append(A)
return T
def mult(A, k):
if not isinstance(A, list) or not (isinstance(k, int)
or isinstance(k, float)):
cm('Either matrix A is not a matrix or k a number', 'fail')
return []
if A == []:
cm('Matrix A is empty', 'fail')
return []
rowsA = len(A)
try:
colsA = len(A[0])
except TypeError:
colsA = 1
# Multiply A by k
Mul = list()
for row in range(rowsA):
Row = list()
for col in range(colsA):
if colsA != 1:
Row.append(k * A[row][col])
else:
Row.append(k * A[row])
if len(Row) == 1:
Mul.append(Row[0])
else:
Mul.append(Row)
return Mul
def NeuralNetwork(x = values, y = outcomes, Print = '', Train = True):
global weights
outcome = u.dot(x, weights)
outcome = u.nonlin(outcome)
if Print == 'test':
if outcome[0] > .5:
cm ("My guess is 1", 'green')
else:
cm ("My guess is 0", 'green')
if Print == 'dev':
cm ('The weights are: {}'.format(weights), 'dev')
cm ('Expected outcomes: {}'.format(y), 'dev')
cm ('Actual outcomes: {}'.format(outcome), 'dev')
err = u.add(y, u.mult(outcome, -1))
#Backpropagation
dw = u.multMatrix(err, u.nonlin(outcome, True))
if Train:
weights = u.add (weights, u.dot(x, dw, t1 = True))
def add(A, B):
if not isinstance(A, list) or not isinstance(B, list):
cm('Either matrix A or B is not a matrix', 'fail')
return []
if A == [] or B == []:
cm('Either matrix A or B is empty', 'fail')
return []
try:
rowsA = len(A)
rowsB = len(B)
colsA = len(A[0])
colsB = len(B[0])
except TypeError:
try:
rowsA += 0
except UnboundLocalError:
rowsA = 1
try:
colsA += 0
except UnboundLocalError:
colsA = 1
try:
rowsB += 0
except UnboundLocalError:
rowsB = 1
try:
colsB += 0
except UnboundLocalError:
colsB = 1
if rowsA != rowsB or colsA != colsB:
cm("Matrix A is not of the same order as matrix B", 'fail')
return []
# Add B to A
Sum = list()
for row in range(rowsA):
Row = list()
for col in range(colsA):
if rowsA != 1 and colsA != 1:
Row.append(A[row][col] + B[row][col])
elif rowsA != 1 and colsA == 1:
Row.append(A[row] + B[row])
elif rowsA == 1 and colsA != 1:
Row.append(A[0][col] + B[0][col])
elif rowsA == 1 and colsA == 1:
A = clean(A)
B = clean(B)
Row.append(A[0] + B[0])
if len(Row) == 1:
Sum.append(Row[0])
else:
Sum.append(Row)
return Sum
def multMatrix(A, B):
if not isinstance(A, list) or not isinstance(B, list):
cm('Either matrix A or B is not a matrix', 'fail')
return []
if A == [] or B == []:
cm('Either matrix A or B is empty', 'fail')
return []
rowsA = len(A)
rowsB = len(B)
try:
colsA = len(A[0])
except TypeError:
colsA = 1
try:
colsB = len(B[0])
except TypeError:
colsB = 1
if rowsA != rowsB or colsA != colsB:
cm("Matrix A is not of the same order as matrix B", 'fail')
return []
# Add B to A
Mult = list()
for row in range(rowsA):
Row = list()
for col in range(colsA):
if rowsA != 1 and colsA != 1:
Row.append(A[row][col] * B[row][col])
elif rowsA != 1 and colsA == 1:
Row.append(A[row] * B[row])
elif rowsA == 1 and colsA != 1:
Row.append(A[0][col] * B[0][col])
elif rowsA == 1 and colsA == 1:
A = clean(A)
B = clean(B)
Row.append(A[0] * B[0])
if len(Row) == 1:
Mult.append(Row[0])
else:
Mult.append(Row)
return Mult
def dot(A, B, t1=False, t2=False):
if not isinstance(A, list) or not isinstance(B, list):
cm('Either matrix A or B is not a matrix', 'fail')
return []
if A == [] or B == []:
cm('Either matrix A or B is empty', 'fail')
return []
if t1:
A = transpose(A)
if t2:
B = transpose(B)
rowsA = len(A)
rowsB = len(B)
try:
colsA = len(A[0])
except TypeError:
colsA = 1
try:
colsB = len(B[0])
except TypeError:
colsB = 1
if colsA != rowsB:
cm("Cols of A does not equal rows of B", 'fail')
return []
dotM = list()
for row in range(rowsA):
R = list()
for c in range(colsB):
sum = 0
for r in range(rowsB):
if colsB != 1:
if rowsA != 1 and colsA != 1:
sum += A[row][r] * B[r][c]
elif rowsA != 1 and colsA == 1:
sum += A[row] * B[0][c]
elif rowsA == 1 and colsA != 1:
sum += A[0][r] * B[r][c]
elif rowsA == 1 and colsA == 1:
A = clean(A)
B = clean(B)
sum += A[0] * B[0][c]
if colsB == 1:
if rowsA != 1 and colsA != 1:
sum += A[row][r] * B[r]
elif rowsA != 1 and colsA == 1:
sum += A[row] * B[0]
elif rowsA == 1 and colsA != 1:
sum += A[0][r] * B[r]
elif rowsA == 1 and colsA == 1:
A = clean(A)
B = clean(B)
sum += A[0] * B[0]
R.append(sum)
if len(R) == 1:
dotM.append(R[0])
else:
dotM.append(R)
return (dotM)