def oneItPerceptron(data_generator, weights, labels):
errors = 0.
num_insts = 0.
# your code here
for sample in data_generator:
feature, label = sample
pred = predict(feature, weights, labels)[0]
num_insts += 1
if pred != label:
errors += 1
for w in feature:
weights[(label, w)] += feature[w]
weights[(pred, w)] -= feature[w]
return weights, errors, num_insts
def oneItPerceptron(data_generator,weights,labels):
errors = 0.
num_insts = 0.
# your code here
for sample in data_generator:
feature, label = sample
pred = predict(feature, weights, labels)[0]
num_insts += 1
if pred != label:
errors += 1
for w in feature:
weights[(label, w)] += feature[w]
weights[(pred, w)] -= feature[w]
return weights, errors, num_insts
def oneItPerceptron(data_generator, weights, labels):
errors = 0.
num_insts = 0.
for instance, label in data_generator:
num_insts += 1
label_pred, scores = predict(instance, weights, labels)
if label_pred != label:
errors += 1
for word, value in instance.iteritems():
weights[(label, word)] += value
weights[(label_pred, word)] -= value
return weights, errors, num_insts
def oneItAvgPerceptron(inst_generator, weights, wsum, labels, Tinit=0):
errors = 0.0
num_insts = 0.0
t = Tinit
for sample in inst_generator:
feature, label = sample
pred = predict(feature, weights, labels)[0]
num_insts += 1
if pred != label:
errors += 1
for w in feature:
weights[(label, w)] += feature[w]
weights[(pred, w)] -= feature[w]
wsum[(label, w)] += t * feature[w]
wsum[(pred, w)] -= t * feature[w]
t += 1
return weights, wsum, errors, num_insts
def oneItAvgPerceptron(inst_generator, weights, wsum, labels, Tinit=0):
errors = 0.
num_insts = 0.
t = Tinit
for sample in inst_generator:
feature, label = sample
pred = predict(feature, weights, labels)[0]
num_insts += 1
if pred != label:
errors += 1
for w in feature:
weights[(label, w)] += feature[w]
weights[(pred, w)] -= feature[w]
wsum[(label, w)] += t * feature[w]
wsum[(pred, w)] -= t * feature[w]
t += 1
return weights, wsum, errors, num_insts
