Killian / decodopr

Blame view

LDA/04b-mmf_mini_ae.py 6.74 KB
edit raw normal view history
7db73861f   Killian   add vae et mmf 
1
2
3
4
5
6
7
8
9
10
11
12
 
  
  # coding: utf-8
  
  # In[2]:
  
  # Import
  import gensim
  from scipy import sparse
  import itertools
  from sklearn import preprocessing
  from keras.models import Sequential
  from keras.optimizers import SGD,Adam
2af8e57f4   Killian   change all 
13
 
  from keras.layers.advanced_activations import ELU,PReLU
7db73861f   Killian   add vae et mmf 
14
15
16
17
18
19
20
21
22
23
24
25
26
27
 
  from mlp import *
  import sklearn.metrics
  import shelve
  import pickle
  from utils import *
  import sys
  import os
  import json
  # In[4]:
  
  infer_model=shelve.open("{}".format(sys.argv[2]))
  in_dir = sys.argv[1]
  #['ASR', 'TRS', 'LABEL']
  # In[6]:
2af8e57f4   Killian   change all 
28
29
30
31
32
 
  if len(sys.argv) > 4 :
      features_key = sys.argv[4]
  else :
      features_key = "LDA"
  save_projection = True
e5108393c   Killian   replace du mlp.p... 
33
34
35
36
 
  json_conf =json.load(open(sys.argv[3]))
  ae_conf = json_conf["ae"]
  
  hidden_size= ae_conf["hidden_size"]
2af8e57f4   Killian   change all 
37
38
39
40
41
42
43
44
45
 
  input_activation = None
  print ae_conf["input_activation"]
  if ae_conf["input_activation"] == "elu":
      print " ELU"
      input_activation = PReLU()
  else:
      print " ELSE"
      input_activation = ae_conf["input_activation"]
  #input_activation=ae_conf["input_activation"]
e5108393c   Killian   replace du mlp.p... 
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
 
  output_activation=ae_conf["output_activation"]
  loss=ae_conf["loss"]
  epochs=ae_conf["epochs"]
  batch=ae_conf["batch"]
  patience=ae_conf["patience"]
  do_do=ae_conf["do"]
  try:
      k = ae_conf["sgd"]
      if ae_conf["sgd"]["name"] == "adam":
          sgd = Adam(lr=ae_conf["sgd"]["lr"])#SGD(lr=0.00001,nesterov=False) #'rmsprop'# Adam(lr=0.00001)#SGD(lr=0.001, momentum=0.9, nesterov=True)
      elif ae_conf["sgd"]["name"] == "sgd":
          sgd = SGD(lr=ae_conf["sgd"]["lr"])
  except: 
      sgd = ae_conf["sgd"]
  
  mlp_conf = json_conf["mlp"]
  mlp_h = mlp_conf["hidden_size"]
  mlp_loss = mlp_conf["loss"]
  mlp_dropouts = mlp_conf["do"]
  mlp_epochs = mlp_conf["epochs"]
  mlp_batch_size = mlp_conf["batch"]
  mlp_input_activation=mlp_conf["input_activation"]
  mlp_output_activation=mlp_conf["output_activation"]
7db73861f   Killian   add vae et mmf 
69
 








e5108393c
 
 Killian
 
replace du  mlp.p...




70
71
72
73
74
75
76
77
78
79
80




  try:
      k = mlp_conf["sgd"]
      if mlp_conf["sgd"]["name"] == "adam":
          mlp_sgd = Adam(lr=mlp_conf["sgd"]["lr"])#SGD(lr=0.00001,nesterov=False) #'rmsprop'# Adam(lr=0.00001)#SGD(lr=0.001, momentum=0.9, nesterov=True)
      elif mlp_conf["sgd"]["name"] == "sgd":
          mlp_sgd = SGD(lr=mlp_conf["sgd"]["lr"])
  except: 
      mlp_sgd = mlp_conf["sgd"]
  
  
  name = json_conf["name"]









7db73861f
 
 Killian
 
add vae et mmf




81
82
83
84
85




  try:
      os.mkdir("{}/{}".format(in_dir,name))
  except:
      pass
  db = shelve.open("{}/{}/ae_model.shelve".format(in_dir,name),writeback=True)









7db73861f
 
 Killian
 
add vae et mmf




86
87




  db["LABEL"]=infer_model["LABEL"]
  #









2af8e57f4
 
 Killian
 
change all




88




  keys = infer_model[features_key].keys()









7db73861f
 
 Killian
 
add vae et mmf




89
90




  
  db["AE"] = {}









2af8e57f4
 
 Killian
 
change all




91




  db[features_key] = {}









7db73861f
 
 Killian
 
add vae et mmf




92




  for mod in keys : 









2af8e57f4
 
 Killian
 
change all




93
94
95
96
97
98
99




      print infer_model[features_key][mod]["TRAIN"].shape
      print infer_model[features_key][mod]["DEV"].shape
      print infer_model[features_key][mod]["TEST"].shape
  
      db[features_key][mod] = train_mlp(infer_model[features_key][mod]["TRAIN"],infer_model["LABEL"][mod]["TRAIN"],
                              infer_model[features_key][mod]["DEV"],infer_model["LABEL"][mod]["DEV"],
                              infer_model[features_key][mod]["TEST"],infer_model["LABEL"][mod]["TEST"],









7db73861f
 
 Killian
 
add vae et mmf




100
101
102




                              mlp_h ,sgd=mlp_sgd,
                              epochs=mlp_epochs,
                              batch_size=mlp_batch_size,









e5108393c
 
 Killian
 
replace du  mlp.p...




103




                              input_activation=mlp_input_activation,









7db73861f
 
 Killian
 
add vae et mmf




104
105
106




                              output_activation=mlp_output_activation,
                              dropouts=mlp_dropouts,
                              fit_verbose=0)









2af8e57f4
 
 Killian
 
change all




107
108




      print input_activation
      res=train_ae(infer_model[features_key][mod]["TRAIN"],infer_model[features_key][mod]["DEV"],infer_model[features_key][mod]["TEST"],









e5108393c
 
 Killian
 
replace du  mlp.p...




109




                   hidden_size,patience = patience,sgd=sgd,









7db73861f
 
 Killian
 
add vae et mmf




110
111
112




                   dropouts=do_do,input_activation=input_activation,output_activation=output_activation,
                   loss=loss,epochs=epochs,batch_size=batch,verbose=0)
      mlp_res_list=[]









2af8e57f4
 
 Killian
 
change all




113
114
115
116
117
118
119
120
121
122
123
124
125




      for nb,layer in enumerate(res) :
          if save_projection:
              pd = pandas.DataFrame(layer[0])
              col_count = (pd.sum(axis=0) != 0)
              pd = pd.loc[:,col_count]
              pd.to_hdf("{}/{}/AE_{}_{}_df.hdf".format(in_dir,name,nb,mod),"TRAIN")
              pd = pandas.DataFrame(layer[1])
              pd = pd.loc[:,col_count]
              pd.to_hdf("{}/{}/AE_{}_{}_df.hdf".format(in_dir,name,nb,mod),"DEV")
              pd = pandas.DataFrame(layer[2])
              pd = pd.loc[:,col_count]
              pd.to_hdf("{}/{}/AE_{}_{}_df.hdf".format(in_dir,name,nb,mod),"TEST")
              del pd









7db73861f
 
 Killian
 
add vae et mmf




126
127
128
129
130




          mlp_res_list.append(train_mlp(layer[0],infer_model['LABEL'][mod]["TRAIN"],
                                        layer[1],infer_model["LABEL"][mod]["DEV"],
                                        layer[2],infer_model["LABEL"][mod]["TEST"],
                                        mlp_h,loss=mlp_loss,dropouts=mlp_dropouts,sgd=mlp_sgd,epochs=mlp_epochs,
                                        output_activation=mlp_output_activation,









e5108393c
 
 Killian
 
replace du  mlp.p...




131




                                        input_activation=mlp_input_activation,









7db73861f
 
 Killian
 
add vae et mmf




132
133




                                        batch_size=mlp_batch_size,fit_verbose=0))
      db["AE"][mod]=mlp_res_list









2af8e57f4
 
 Killian
 
change all




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




  if "ASR" in keys and "TRS" in keys:
      mod = "ASR"
      mod2= "TRS"
      mlp_res_list=[]
  
      res = train_ae(infer_model[features_key][mod]["TRAIN"],
                      infer_model[features_key][mod]["DEV"],
                      infer_model[features_key][mod]["TEST"],
                      hidden_size,dropouts=do_do,patience = patience,
                      sgd=sgd,input_activation=input_activation,output_activation=output_activation,loss=loss,epochs=epochs,
                      batch_size=batch,
                      y_train=infer_model[features_key][mod]["TRAIN"],
                      y_dev=infer_model[features_key][mod2]["DEV"],
                      y_test=infer_model[features_key][mod2]["TEST"])
  
      for nb,layer in enumerate(res) :
          if save_projection:
              pd = pandas.DataFrame(layer[0])
              col_count= (pd.sum(axis=0) != 0)
              pd = pd.loc[:,col_count]
              pd.to_hdf("{}/{}/AE_{}_{}_df.hdf".format(in_dir,name,nb,"SPE"),"TRAIN")
              pd = pandas.DataFrame(layer[1])
              pd = pd.loc[:,col_count]
              pd.to_hdf("{}/{}/AE_{}_{}_df.hdf".format(in_dir,name,nb,"SPE"),"DEV")
              pd = pandas.DataFrame(layer[2])
              pd = pd.loc[:,col_count]
              pd.to_hdf("{}/{}/AE_{}_{}_df.hdf".format(in_dir,name,nb,"SPE"),"TEST")
              del pd
  
          mlp_res_list.append(train_mlp(layer[0],infer_model["LABEL"][mod]["TRAIN"],
                                        layer[1],infer_model["LABEL"][mod]["DEV"],
                                        layer[2],infer_model["LABEL"][mod]["TEST"],
                                        mlp_h,loss=mlp_loss,dropouts=mlp_dropouts,sgd=mlp_sgd,epochs=mlp_epochs,
                                        output_activation=mlp_output_activation,
                                        input_activation=input_activation,
                                        batch_size=mlp_batch_size,fit_verbose=0))
  
      db["AE"]["SPE"] = mlp_res_list









7db73861f
 
 Killian
 
add vae et mmf




172
173
174




  
  db.sync()
  db.close()