From 93b9b214dd95858bddec508e991847a3cab00333 Mon Sep 17 00:00:00 2001
From: Nikolai Hartmann <Nikolai.Hartmann@physik.uni-muenchen.de>
Date: Mon, 10 Sep 2018 11:12:54 +0200
Subject: [PATCH] attempt to modify training to always use fit_generator +
option for KFold cross validation
---
toolkit.py | 114 +++++++++++++++++++++++++++++------------------------
1 file changed, 62 insertions(+), 52 deletions(-)
diff --git a/toolkit.py b/toolkit.py
index f108d1d..51efd3f 100755
--- a/toolkit.py
+++ b/toolkit.py
@@ -33,6 +33,7 @@ from sklearn.preprocessing import StandardScaler, RobustScaler
from sklearn.externals import joblib
from sklearn.metrics import roc_curve, auc
from sklearn.utils.extmath import stable_cumsum
+from sklearn.model_selection import KFold
from keras.models import Sequential, Model, model_from_json
from keras.layers import Dense, Dropout, Input, Masking, GRU, LSTM, concatenate, SimpleRNN
from keras.callbacks import History, EarlyStopping, CSVLogger, ModelCheckpoint, TensorBoard
@@ -125,6 +126,10 @@ class ClassificationProject(object):
:param validation_split: split off this fraction of training events for loss evaluation
+ :param kfold_splits: if given, split into this number of of subsets to perform KFold cross validation
+
+ :param kfold_index: index of the subset to leave out for kfold validation
+
:param activation_function: activation function in the hidden layers
:param activation_function_output: activation function in the output layer
@@ -224,6 +229,8 @@ class ClassificationProject(object):
dropout_input=None,
batch_size=128,
validation_split=0.33,
+ kfold_splits=None,
+ kfold_index=0,
activation_function='relu',
activation_function_output='sigmoid',
scaler_type="WeightedRobustScaler",
@@ -284,6 +291,8 @@ class ClassificationProject(object):
self.dropout_input = dropout_input
self.batch_size = batch_size
self.validation_split = validation_split
+ self.kfold_splits = kfold_splits
+ self.kfold_index = kfold_index
self.activation_function = activation_function
self.activation_function_output = activation_function_output
self.scaler_type = scaler_type
@@ -347,6 +356,7 @@ class ClassificationProject(object):
self._model = None
self._history = None
self._callbacks_list = []
+ self._train_val_idx = None
# track the number of epochs this model has been trained
self.total_epochs = 0
@@ -842,20 +852,53 @@ class ClassificationProject(object):
@property
def validation_data(self):
- "Validation data. Attention: Shuffle training data before using this!"
- if not self.data_shuffled:
- raise ValueError("Training data isn't shuffled, can't split of validation data")
- split_index = int((1-self.validation_split)*len(self.x_train))
- return self.x_train[split_index:], self.y_train[split_index:], self.w_train_tot[split_index:]
+ "Validation data"
+ idx = self.train_val_idx[1]
+ return self.x_train[idx], self.y_train[idx], self.w_train_tot[idx]
@property
def training_data(self):
- "Training data with validation data split off. Attention: Shuffle training data before using this!"
- if not self.data_shuffled:
- raise ValueError("Training data isn't shuffled, can't split of validation data")
- split_index = int((1-self.validation_split)*len(self.x_train))
- return self.x_train[:split_index], self.y_train[:split_index], self.w_train_tot[:split_index]
+ "Training data with validation data split off"
+ idx = self.train_val_idx[0]
+ return self.x_train[idx], self.y_train[idx], self.w_train_tot[idx]
+
+
+ @property
+ def train_val_idx(self):
+ if self._train_val_idx is None:
+ if self.kfold_splits is not None:
+ kfold = KFold(self.kfold_splits, shuffle=True, random_state=self.shuffle_seed)
+ for i, train_val_idx in kfold.split(self.x_train):
+ if i == self.kfold_index:
+ self._train_val_idx = train_val_idx
+ break
+ else:
+ raise IndexError("Index {} out of range for kfold (requested {} splits)".format(self.kfold_index, self.kfold_splits))
+ else:
+ split_index = int((1-self.validation_split)*len(self.x_train))
+ np.random.seed(self.shuffle_seed)
+ shuffled_idx = np.random.permutation(len(self.x_train))
+ self._train_val_idx = (shuffled_idx[:split_index], shuffled_idx[split_index:])
+ return self._train_val_idx
+
+
+ @property
+ def steps_per_epoch(self):
+ return int(float(len(self.train_val_idx[0]))/float(self.batch_size))
+
+
+ def yield_batch(self):
+ x_train, y_train, w_train = self.x_train, self.y_train, self.w_train_tot
+ train_idx = list(self.train_val_idx[0])
+ np.random.seed(self.shuffle_seed+1)
+ while True:
+ shuffled_idx = np.random.permutation(train_idx)
+ for start in range(0, len(shuffled_idx), int(self.batch_size)):
+ x_batch = x_train[shuffled_idx[start:start+int(self.batch_size)]]
+ y_batch = y_train[shuffled_idx[start:start+int(self.batch_size)]]
+ w_batch = w_train[shuffled_idx[start:start+int(self.batch_size)]]
+ yield (x_batch, y_batch, w_batch)
def yield_single_class_batch(self, class_label):
@@ -894,8 +937,6 @@ class ClassificationProject(object):
self.load()
- self.shuffle_training_data()
-
for branch_index, branch in enumerate(self.fields):
self.plot_input(branch_index)
@@ -905,18 +946,11 @@ class ClassificationProject(object):
if not self.balance_dataset:
try:
self.is_training = True
- np.random.seed(self.shuffle_seed+1) # since we use keras shuffling here
- self.model.fit(self.x_train,
- # the reshape might be unnescessary here
- self.y_train.reshape(-1, 1),
- epochs=epochs,
- validation_split=self.validation_split,
- # we have to multiply by class weight since keras ignores class weight if sample weight is given
- # see https://github.com/keras-team/keras/issues/497
- sample_weight=self.w_train_tot,
- shuffle=True,
- batch_size=self.batch_size,
- callbacks=self.callbacks_list)
+ self.model.fit_generator(self.yield_batch(),
+ steps_per_epoch=self.steps_per_epoch,
+ epochs=epochs,
+ validation_data=self.validation_data,
+ callbacks=self.callbacks_list)
self.is_training = False
except KeyboardInterrupt:
logger.info("Interrupt training - continue with rest")
@@ -1761,32 +1795,6 @@ class ClassificationProjectRNN(ClassificationProject):
return self._model
- def train(self, epochs=10):
- self.load()
-
- self.shuffle_training_data()
-
- for branch_index, branch in enumerate(self.fields):
- self.plot_input(branch_index)
-
- self.total_epochs = self._read_info("epochs", 0)
-
- try:
- self.is_training = True
- logger.info("Training on batches for RNN")
- # note: the batches have class_weight already applied
- self.model.fit_generator(self.yield_batch(),
- steps_per_epoch=int(len(self.training_data[0])/self.batch_size),
- epochs=epochs,
- validation_data=self.validation_data,
- callbacks=self.callbacks_list)
- self.is_training = False
- except KeyboardInterrupt:
- logger.info("Interrupt training - continue with rest")
-
- self.checkpoint_model()
-
-
def clean_mask(self, x):
"""
Mask recurrent fields such that once a masked value occurs,
@@ -1846,9 +1854,11 @@ class ClassificationProjectRNN(ClassificationProject):
def yield_batch(self):
- x_train, y_train, w_train = self.training_data
+ x_train, y_train, w_train = self.x_train, self.y_train, self.w_train_tot
+ train_idx = list(self.train_val_idx[0])
+ np.random.seed(self.shuffle_seed+1)
while True:
- shuffled_idx = np.random.permutation(len(x_train))
+ shuffled_idx = np.random.permutation(train_idx)
for start in range(0, len(shuffled_idx), int(self.batch_size)):
x_batch = x_train[shuffled_idx[start:start+int(self.batch_size)]]
y_batch = y_train[shuffled_idx[start:start+int(self.batch_size)]]
--
GitLab