def globallossfunction(tr, sub):
threshold = 10
loss = np.zeros((len(tr), len(sub)))
distance = np.zeros((len(tr), len(sub)))
for jj in range(len(tr)):
for ii in range(len(sub)):
loss[jj, ii], distance[jj, ii] = lossfunction(tr[jj], sub[ii])
#loss[jj, ii], distance[jj, ii] = lossfunction__old1(tr[jj], sub[ii])
print loss
return loss, distance, threshold
def globallossfunction(tr, sub):
threshold = 10
loss = np.zeros((len(tr), len(sub)))
distance = np.zeros((len(tr), len(sub)))
for jj in range(len(tr)):
for ii in range(len(sub)):
loss[jj, ii], distance[jj, ii] = lossfunction(tr[jj], sub[ii])
#loss[jj, ii], distance[jj, ii] = lossfunction__old1(tr[jj], sub[ii])
print loss
return loss, distance, threshold
def globallossfunction(tr, sub):
threshold = 80
loss = np.zeros((len(tr), len(sub)))
for jj in range(len(tr)):
for ii in range(len(sub)):
loss[jj, ii] = lossfunction(tr[jj], sub[ii])
return loss, threshold
def resample(self):
if np.sum(self.prob) > 0:
p = self.p
y, x = p.shape
new_p = np.zeros((y, x))
for ii in range(len(self.p)):
idx = self.pmfrnd()
new_p[ii, :] = p[idx, :]
self.p = new_p
def resample(self):
if np.sum(self.prob) > 0:
p = self.p
y, x = p.shape
new_p = np.zeros((y, x))
for ii in range(len(self.p)):
idx = self.pmfrnd()
new_p[ii, :] = p[idx, :]
self.p = new_p