02b-transfert_ae.py
3.46 KB
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
# 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
from keras.layers.advanced_activations import ELU,PReLU
from keras.callbacks import ModelCheckpoint
from mlp import *
import pandas as pd
import sklearn.metrics
from sklearn.preprocessing import LabelBinarizer
import shelve
import pickle
from utils import *
import sys
import os
import json
# In[4]:
in_dir = sys.argv[1]
#['ASR', 'TRS', 'LABEL']
# In[6]:
json_conf =json.load(open(sys.argv[2]))
mlp_conf = json_conf["transfert"]
hidden_size = mlp_conf["hidden_size"]
loss = mlp_conf["loss"]
patience = mlp_conf["patience"]
dropouts = mlp_conf["do"]
epochs = mlp_conf["epochs"]
batch_size = mlp_conf["batch"]
input_activation=mlp_conf["input_activation"]
output_activation=mlp_conf["output_activation"]
try:
k = mlp_conf["sgd"]
if mlp_conf["sgd"]["name"] == "adam":
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":
sgd = SGD(lr=mlp_conf["sgd"]["lr"])
except:
sgd = mlp_conf["sgd"]
name = json_conf["name"]
#
proj_hdf = pandas.HDFStore("{}/{}/MLP_proj_df.hdf".format(in_dir,name))
hdf_keys = proj_hdf.keys()
proj_hdf.close()
hdf_mods = set([ x.split("/")[1] for x in hdf_keys ])
hdf_lvl = set( [ x.split("/")[2] for x in hdf_keys ])
hdf_crossval = set([ x.split("/")[3] for x in hdf_keys ])
print hdf_mods
print hdf_lvl
print hdf_crossval
hdf_proj_path = "{}/{}/MLP_proj_df.hdf".format(in_dir,name)
transfert_proj_path = "{}/{}/transfert_proj_df.hdf".format(in_dir,name)
mod1,mod2 = "ASR","TRS"
for lvl in hdf_lvl :
x_train_ASR = pandas.read_hdf(hdf_proj_path,key="/{}/{}/{}".format(mod1,lvl,"TRAIN"))
x_dev_ASR = pandas.read_hdf(hdf_proj_path,key="/{}/{}/{}".format(mod1,lvl,"DEV"))
x_test_ASR = pandas.read_hdf(hdf_proj_path,key="/{}/{}/{}".format(mod1,lvl,"TEST"))
x_train_TRS = pandas.read_hdf(hdf_proj_path,key="/{}/{}/{}".format(mod2,lvl,"TRAIN"))
x_dev_TRS = pandas.read_hdf(hdf_proj_path,key="/{}/{}/{}".format(mod2,lvl,"DEV"))
x_test_TRS = pandas.read_hdf(hdf_proj_path,key="/{}/{}/{}".format(mod2,lvl,"TEST"))
if x_train_ASR.shape[1] <= 8 :
continue
pred = train_ae(x_train_ASR.values,
x_dev_ASR.values,
x_test_ASR.values,
hidden_size ,sgd=sgd,
y_train=x_train_TRS.values,
y_dev=x_dev_TRS.values,
y_test=x_test_TRS.values,
epochs=epochs,
patience=patience,
batch_size=batch_size,
input_activation=input_activation,
output_activation=output_activation,
dropouts=dropouts,
best_mod=True,
verbose=1)
for num_layer,layer in enumerate(pred):
transfert_train = pd.DataFrame(layer[0])
transfert_dev = pd.DataFrame(layer[1])
transfert_test = pd.DataFrame(layer[2])
transfert_train.to_hdf(transfert_proj_path,"{}/{}/TRAIN".format(lvl,"layer"+str(num_layer)))
transfert_dev.to_hdf(transfert_proj_path,"{}/{}/DEV".format(lvl,"layer"+str(num_layer)))
transfert_test.to_hdf(transfert_proj_path,"{}/{}/TEST".format(lvl,"layer"+str(num_layer)))