-
Notifications
You must be signed in to change notification settings - Fork 0
/
bagg.py
206 lines (198 loc) · 10 KB
/
bagg.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
from microtc.textmodel import TextModel
import json
import pandas as pd
from sklearn.preprocessing import MinMaxScaler, StandardScaler, Normalizer
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelEncoder
from sklearn.svm import LinearSVC
from sklearn.naive_bayes import GaussianNB
from sklearn.naive_bayes import MultinomialNB
from sklearn.linear_model import RidgeClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import f1_score, precision_score, recall_score
from sklearn.neighbors import KNeighborsClassifier
from sklearn.neural_network import MLPClassifier
import numpy as np
from scipy.optimize import minimize, shgo, differential_evolution
from flair.embeddings import (CharacterEmbeddings, DocumentPoolEmbeddings, BytePairEmbeddings,TransformerWordEmbeddings,
DocumentRNNEmbeddings,BytePairEmbeddings,DocumentLSTMEmbeddings,FlairEmbeddings)
from flair.data import Sentence
from joblib import Parallel, delayed
#ml='models/malayalam_params.json'
#l=json.load(open(ml))
#lm={}
#mlf='data/malayalam_train.json'
#mx=pd.read_json(mlf, lines=True)
#X=mx.text.values
#Y=mx.klass.values
#le = LabelEncoder()
#le.fit(Y)
#y=le.transform(Y)
#Xt,Xv,yt,yv=train_test_split(X,y,test_size=0.2,random_state=33, stratify=y)
class bagginTextModels():
def _train_model(self,model):
xt=model.get('xt')
print(f"Training model : {model['name']}, {xt.shape}")
clf=LinearSVC()
#clf=model['name']=='mtc' and LinearSVC() or RidgeClassifier()#RandomForestClassifier()
if 'xv'in model.keys() :
xv=model.get('xv')
model['clf']=clf.fit(xt,self.yt)#LinearSVC(max_iter=5000).fit(xt,self.yt)
#model['clf']=RandomForestClassifier().fit(xt,self.yt)
#if model['name']=='mtc':
yp=model['clf'].decision_function(xv)
#else:
# yp=model['clf'].predict_proba(xv)
yp=Normalizer().fit_transform(yp)
model['macroF1']=f1_score(self.yv,np.argmax(yp,axis=1),average='macro')
model['weightedF1']=f1_score(self.yv,np.argmax(yp,axis=1),average='weighted')
model['probas']=yp
### Fit model with all avaliable data
else:
model['clf']=clf.fit(xt,self.y)
return model
def __create_models(self):
models=[]
models_fit=[]
#for _params in self.model_params:
_params={}
for k,v in self.params.items():
if k.startswith('_'):
continue
_params[k]=v
self.textModels=dict(mtc=TextModel(_params).fit(self.train),
#charEmb=DocumentPoolEmbeddings([CharacterEmbeddings()]),
#charLangEmb=DocumentPoolEmbeddings([CharacterEmbeddings(),BytePairEmbeddings(self.lang)]),
##charMultiEmb=DocumentPoolEmbeddings([CharacterEmbeddings(),BytePairEmbeddings('multi')]),
langEmb=DocumentPoolEmbeddings([BytePairEmbeddings(self.lang)]),
charLangMultiEmb=DocumentPoolEmbeddings([CharacterEmbeddings(),BytePairEmbeddings(self.lang),
BytePairEmbeddings('multi')]),
langMultiEmb=DocumentPoolEmbeddings([BytePairEmbeddings(self.lang),BytePairEmbeddings('multi')]),
bytePairEMB=DocumentPoolEmbeddings([BytePairEmbeddings('multi')]),
#flairEmbF=DocumentPoolEmbeddings([FlairEmbeddings('multi-forward')]),
#flairEmbB=DocumentPoolEmbeddings([FlairEmbeddings('multi-backward')]),
#bertEMB=DocumentPoolEmbeddings([TransformerWordEmbeddings('bert-base-uncased', layers='-1')])
)
for km,tmodel in self.textModels.items():
models.append({'name':km})
models_fit.append({'name':km})
if km=='mtc':
xt=tmodel.transform(self.train)
xv=tmodel.transform(self.validation)
X=tmodel.transform(self.data)
else:
sentences_train=[Sentence(txt) for txt in self.train]
tmodel.embed(sentences_train)
xt=np.array([e.get_embedding().cpu().detach().numpy() for e in sentences_train])
sentences_val=[Sentence(txt) for txt in self.validation]
tmodel.embed(sentences_val)
xv=np.array([e.get_embedding().cpu().detach().numpy() for e in sentences_val])
sentences=[Sentence(txt) for txt in self.data]
tmodel.embed(sentences)
X=np.array([e.get_embedding().cpu().detach().numpy() for e in sentences])
models[-1]['xv']=xv
models[-1]['xt']=xt
models_fit[-1]['xt']=X
#max_iter=5000
#if km=='mtc': max_iter=1000
#if km=='langMulti': max_iter=5000
#self.models[-1]['clf']=LinearSVC(max_iter=max_iter).fit(xt,self.yt)
#yp=self.models[-1]['clf'].decision_function(xv)
#scaler=Normalizer().fit(yp)
#self.models[-1]['macroF1']=f1_score(self.yv,np.argmax(scaler.transform(yp),axis=1),average='weighted')
#self.models[-1]['weightedF1']=f1_score(self.yv,np.argmax(scaler.transform(yp),axis=1),average='weighted')
#self.models[-1]['score']=f1_score(self.yv,np.argmax(yp,axis=1),average='weighted')
#self.models[-1]['probas']=scaler.transform(yp)
### Fit model with all avaliable data
#self.models_fit[-1]['clf']=LinearSVC(max_iter=max_iter).fit(X,self.y)
print('Fitting Ensemble')
#self.models = Parallel(n_jobs=5)(delayed(self._train_model)(md) for md in models)
#self.models_fit = Parallel(n_jobs=5)(delayed(self._train_model)(md) for md in models_fit)
self.models,self.models_fit=[],[]
for md,mdf in zip(models, models_fit):
self.models.append(self._train_model(md))# = [self._train_model(md) for md in models]
self.models_fit.append(self._train_model(md))
def __init__(self,data,labels,model_params, test_data=None, test_labels=None,test_size=0.2, lang='ml'):
self.lang=lang
self.labels=labels
self.params=model_params
self.LabelEncoder = LabelEncoder()
self.LabelEncoder.fit(labels)
self.y=self.LabelEncoder.transform(labels)
if test_data is None:
self.data=data
self.train,self.validation,self.yt,self.yv=train_test_split(self.data,self.y,test_size=0.2,
random_state=33, stratify=self.y)
else:
self.data=np.concatenate((data,test_data),axis=0)
self.labels=np.concatenate((labels,test_labels))
self.train,self.validation,self.yt,self.yv=data,test_data,self.y,self.LabelEncoder.transform(test_labels)
self.y=self.LabelEncoder.transform(self.labels)
self.__create_models()
def dotF1(self,alphas):
probas=alphas[0]*self.models[0]['probas']
for alpha,model in zip(alphas[1:],self.models[1:]):
probas=probas+alpha*model['probas']
yp=np.argmax(probas,axis=1)
#return -f1_score(self.yv,yp,average='weighted')
return -f1_score(self.yv,yp,average='weighted')#*f1_score(self.yv,yp,average='macro')
def one_constraint(self,alphas):
return 1-alphas.sum()
def optimize(self):
cons=[{'type':'ineq','fun': self.one_constraint}]
n_models=len(self.models)
#alphas0=np.array([1/n_models for i in range(n_models)])
scores=np.array([model['macroF1'] for model in self.models])
alphas0=scores/scores.sum()
#print("x0",alphas0)
bnds=[(0.0,1.0) for alpha in alphas0]
#sol=minimize(self.dotF1,alphas0,method='SLSQP',constraints=cons,bounds=bnds)
sol=differential_evolution(self.dotF1,bnds)
#sol=shgo(self.dotF1,constraints=cons,bounds=bnds)
self.sol=sol
self.weights=sol.x
def predict(self,X):
#sentences=[Sentence(txt) for txt in X]
ft=True
Yp=[]
for w, model in zip(self.weights,self.models_fit):
if model['name']=='mtc':
x=self.textModels['mtc'].transform(X)
else:
sentences=[Sentence(txt) for txt in X]
self.textModels[model['name']].embed(sentences)
x=np.array([e.get_embedding().cpu().detach().numpy() for e in sentences])
#if model['name']=='mtc':
yp=model['clf'].decision_function(x)
#else:
# yp=model['clf'].predict_proba(x)
if ft:
Yp=w*Normalizer().fit_transform(yp)
ft=False
else:
Yp=Yp+w*Normalizer().fit_transform(yp)
return list(np.argmax(Yp,axis=1))
import pickle,gc
if __name__=='__main__':
from bagg import bagginTextModels
for lang,desc in [('ml','malayalam'),('ta','tamil')]:
#for lang,desc in [('ta','tamil')]:
gc.collect()
tl=f'models/{desc}_params.json'
tp=json.load(open(tl))
tlf=f'data/{desc}_train.json'
data=pd.read_json(tlf, lines=True)
tX=data.text.values
tY=data.klass.values
tdata=pd.read_json(f'data/{desc}_dev.json', lines=True)
xtt=[txt for txt in tdata.text.values]
ytt=[txt for txt in tdata.klass.values]
#bm=bagginTextModels(tX,tY,tp[0],xtt,ytt,lang=lang)
bm=bagginTextModels(tX,tY,tp[0],lang=lang)
bm.optimize()
ypp=bm.predict(xtt)
[print((model['name'], model['macroF1'], model['weightedF1'])) for model in bm.models]
print('pred',precision_score(bm.LabelEncoder.transform(ytt),ypp, average='weighted'))
print('rec',recall_score(bm.LabelEncoder.transform(ytt),ypp, average='weighted'))
print('Wieghted',f1_score(bm.LabelEncoder.transform(ytt),ypp, average='weighted'))
pickle.dump(bm,open(f'{desc}_model_dev_final.pk','wb'))