def aPosterioriDecoding(seq, fileNameEmis, fileNameTrans):
statesGraph = graph.createGraph(fileNameEmis,
fileNameTrans) # it's a dictionary
f = forward(seq, fileNameEmis, fileNameTrans)
b = backward(seq, fileNameEmis, fileNameTrans)
seq = seq[1:-1]
psm = f[len(f) - 1][len(f[0]) - 1]
pi = matrix.createMatrix(len(statesGraph["A"]["A", "States"]), len(seq),
0.)
print "F:"
matrix.printMatrix(f)
print "B:"
matrix.printMatrix(b)
print "Seq:", seq
print "Size of Seq: ", len(seq)
for i in range(len(seq)): #pi(i) -> position in the sequence = collunms
scores = []
for k in range(1, len(statesGraph["A"][
"A", "States"])): #num of states that produce chars = lines
scores.append((f[k][i + 1] * b[k][i + 1]) / psm)
pi[0][i] = scores.index(max(scores)) + 1 #states are from 1 to lines-1
pi[1][i] = scores[0]
pi[2][i] = scores[1]
del (scores)
return pi
def main(argv):
inputfile1 = ''
inputfile2 = ''
print(len(argv) / 2) - 1
try:
inputfileEmission = argv[0]
inputfileTransmission = argv[1]
seqfile = argv[2]
seq = readSequence(seqfile)
print "seqfile: " + str(seqfile)
process = str(seqfile[len(seqfile) - 1])
print "Process: " + process
print 'Input file is "', inputfile1
print 'Input file is "', inputfile2
# To test, please, uncomment what you want to test:
filename = "output.txt" + process
#filename = "output.txt"
stringtowrite = "method: aPosterioriDecoding \n"
#m = forward(seq, inputfileEmission, inputfileTransmission)
#print backward(seq, inputfileEmission, inputfileTransmission)
#trace = viterbi(seq, inputfileEmission, inputfileTransmission)
#print trace
#print giveViterbiTrace(trace)
apd = aPosterioriDecoding(seq, inputfileEmission,
inputfileTransmission)
matrix.printMatrix(apd)
stringtowrite += matrix.stringToMatrix(apd)
writeFile(filename, stringtowrite)
except getopt.GetoptError:
print 'HMM.py <inputfile1> <inputfile2> -o <outputfile>'
sys.exit(2)
def forward(seq, filenameEmission, filenameTransmission):
forwardGraph = graph.createGraph(filenameEmission, filenameTransmission)
#print forwardGraph
char = seq[1]
states = forwardGraph["E"]["E", "States"]
statesCompl = states[:]
statesCompl.insert(0, "0")
statesCompl.append("0")
statesCompl = ["0", "Y", "N", "0"]
print statesCompl
m = matrix.createMatrix(len(states) + 2, len(seq), 0)
# intialisation part
matrix.atualiseValueIntoMatrix(m, 0, 0, 1)
# recurrence part
# recurrence 1
for i in range(1, len(m) - 1):
m[i][1] = float(forwardGraph["A"]["0", statesCompl[i]]) * float(
forwardGraph["E"][statesCompl[i], char])
# recurrence 2
for j in range(2, len(m[0]) - 1): #jump the first 2 collumns
char = seq[j]
for i in range(1, len(m) - 1): # jump the begin and end states
for l in range(1, len(m) - 1):
#print "L:",statesCompl[l],", K:",statesCompl[i]
m[i][j] += float(m[l][j - 1]) * float(
forwardGraph["A"][statesCompl[l], statesCompl[i]]) * float(
forwardGraph["E"][statesCompl[i], char])
'''print "partial value: ",float(m[i][j])
print "lines value:", float(m[l][j-1])
print "AKL: ",float(forwardGraph["A"][statesCompl[l],statesCompl[i]])
print "Elc: ", float(forwardGraph["E"][statesCompl[i],char])'''
for i in range(1, len(m) - 1):
m[len(m) - 1][len(m[0]) - 1] += float(m[i][len(m[0]) - 2]) * float(
forwardGraph["A"][statesCompl[i], "0"])
matrix.printMatrix(m)
return m
def main():
"""
path=input("Enter file name with path(eg. D:\lecture1\data.txt):")
base=input("Enter the number of polynomial bases:")
l=input("Enter lambda:")
=ReadFile(path)
base=int(base)
l=float(l)
"""
d = ReadFile('D:\桌面用\MeachinLearning\MLclass\lecture1\data.txt')
#d=ReadFile('data.txt')
base = 3
l = 0.5
print("Problem A.")
A, B = FindAandB(d, base)
#matrix.printMatrix(A,"A")
#matrix.printMatrix(B,"B")
AT = matrix.TransposeMatrix(A)
#matrix.printMatrix(AT,"AT")
ATA = matrix.Mul(AT, A)
#matrix.printMatrix(ATA,"ATA")
eye = matrix.eyeMatrix(base, l)
_ATA = matrix.Add(ATA, eye)
#matrix.printMatrix(_ATA,"ATA+λi")
L, U = matrix.FindLU(_ATA)
#matrix.printMatrix(L,"L")
#matrix.printMatrix(U,"U")
ATAi = matrix.FindInverseFromLU(L, U)
print("1.")
matrix.printMatrix(ATAi, "Inverse of ATA+λi")
print("2.")
_x = matrix.Mul(matrix.Mul(ATAi, AT), B)
#matrix.printMatrix(_x,"_x")
printEquation(_x)
error = matrix.Sub(matrix.Mul(A, _x), B)
w_error = matrix.Mul(matrix.TransposeMatrix(_x), _x)
w_error = matrix.Factor(w_error, l)
#matrix.printMatrix(w_error,"weight error")
value_error = matrix.Mul(matrix.TransposeMatrix(error), error)
#matrix.printMatrix(value_error,"value error")
matrix.printMatrix(matrix.Add(value_error, w_error), "a_error")
print("------------------------")
#------------------------------------------------------------------
print("Problem b.")
h = HessianMatrix(d, base)
#matrix.printMatrix(h,"Hessian Matrix")
hL, hU = matrix.FindLU(h)
hi = matrix.FindInverseFromLU(hL, hU)
# Random generate the coefficient
_x0 = [[random.random()] for i in range(base)]
#matrix.printMatrix(_x0,"_x0")
g = GenerateGradientMatrix(_x0, d, base)
#matrix.printMatrix(g,"Gradient")
print("1.")
_x0 = matrix.Sub(_x0, matrix.Mul(hi, g))
#matrix.printMatrix(_x0,"solution")
printEquation(_x0)
error = matrix.Sub(matrix.Mul(A, _x0), B)
value_error = matrix.Mul(matrix.TransposeMatrix(error), error)
matrix.printMatrix(value_error, "b_error")
arr[ci][cj-1] = 0
seen[(ci, cj-1)] = True
if ci < len(arr)-1 and (ci+1, cj) not in seen and arr[ci+1][cj] == 1:
queue.append((ci+1, cj))
arr[ci+1][cj] = 0
seen[(ci+1, cj)] = True
if cj < len(arr[ci])-1 and (ci, cj+1) not in seen and arr[ci][cj+1] == 1:
queue.append((ci, cj + 1))
arr[ci][cj+1] = 0
seen[(ci, cj+1)] = True
arr = [
[1, 1, 1, 0],
[1, 1, 1, 0],
[1, 0, 0, 0],
[1, 0, 0, 0],
[1, 0, 0, 0],
[1, 0, 0, 0],
[1, 0, 0, 0],
[1, 0, 0, 0],
]
expectedResult = 2
s = Solution()
printMatrix(arr)
matrix.printMatrix(b)
print "Seq:", seq
print "Size of Seq: ", len(seq)
for i in range(len(seq)): #pi(i) -> position in the sequence = collunms
scores = []
for k in range(1, len(statesGraph["A"][
"A", "States"])): #num of states that produce chars = lines
scores.append((f[k][i + 1] * b[k][i + 1]) / psm)
pi[0][i] = scores.index(max(scores)) + 1 #states are from 1 to lines-1
pi[1][i] = scores[0]
pi[2][i] = scores[1]
del (scores)
return pi
# To test, please, uncomment what you want to test:
filename = "output.txt"
stringtowrite = "method: aPosterioriDecoding \n"
#m = forward(seq, "probabilityEmission.txt", "probabilityTransmission.txt")
#print backward(seq, "probabilityEmission.txt", "probabilityTransmission.txt")
#trace = viterbi(seq, "probabilityEmission.txt", "probabilityTransmission.txt")
#print trace
#print giveViterbiTrace(trace)
apd = aPosterioriDecoding(seq, "probabilityEmission.txt",
"probabilityTransmission.txt")
matrix.printMatrix(apd)
stringtowrite += matrix.stringToMatrix(apd)
writeFile(filename, stringtowrite)