I am working on a image classification project and my model doesn't seem to train properly.
My dataset is made of 4000 images each with a shape of (120,120,3).
Test set represents 20% of the total dataset.
All images have been correctly labeled.
The images are normalized and one-hot encoded. For now I use only two targets, but I will add one more one I start getting decent results.
I use a batch size of 16
I want to use a CNN model.
My current model :
model = keras.models.Sequential()
model.add(Conv2D(filters=16, kernel_size=(6,6), input_shape=(IMG_SIZE,IMG_SIZE,3), activation='relu',))
model.add(MaxPool2D(pool_size=(2,2)))
model.add(Dropout(0.2))
model.add(Conv2D(filters=32, kernel_size=(5,5), activation='relu',))
model.add(MaxPool2D(pool_size=(2,2)))
model.add(Dropout(0.2))
model.add(Conv2D(filters=64, kernel_size=(4,4), activation='relu',))
model.add(MaxPool2D(pool_size=(2,2)))
model.add(Dropout(0.2))
model.add(Conv2D(filters=128, kernel_size=(3,3), activation='relu',))
model.add(MaxPool2D(pool_size=(2,2)))
model.add(Dropout(0.2))
model.add(Conv2D(filters=256, kernel_size=(2,2), activation='relu',))
model.add(MaxPool2D(pool_size=(2,2)))
model.add(Dropout(0.2))
model.add(Flatten())
model.add(Dense(64, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(64, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(2, activation='softmax'))
model.compile(optimizer='nadam',
loss='categorical_crossentropy',
metrics=['accuracy'])
model.summary()
model summary gives :
Total params: 273,330
Trainable params: 273,330
Non-trainable params: 0
from tensorflow.keras.callbacks import EarlyStopping
early_stop = EarlyStopping(monitor='val_loss',patience=10)
history = model.fit(x_train_sample, y_train_sample,
batch_size = BATCH_SIZE,
epochs = EPOCHS,
verbose = 1,
validation_data = (x_test, y_test)
,callbacks=[early_stop,PlotLossesKeras()])
When I run my model for 30 epochs, earlystopping triggers.
Epoch 1/30
43/43 [==============================] - 9s 205ms/step - loss: 0.1109 - accuracy: 0.9531 - val_loss: 0.5259 - val_accuracy: 0.8397
Epoch 2/30
43/43 [==============================] - 10s 231ms/step - loss: 0.0812 - accuracy: 0.9692 - val_loss: 0.5793 - val_accuracy: 0.8355
Epoch 3/30
43/43 [==============================] - 9s 219ms/step - loss: 0.1000 - accuracy: 0.9721 - val_loss: 0.5367 - val_accuracy: 0.8547
Epoch 4/30
43/43 [==============================] - 9s 209ms/step - loss: 0.0694 - accuracy: 0.9707 - val_loss: 0.6101 - val_accuracy: 0.8269
Epoch 5/30
43/43 [==============================] - 9s 203ms/step - loss: 0.0891 - accuracy: 0.9633 - val_loss: 0.6116 - val_accuracy: 0.8419
Epoch 6/30
43/43 [==============================] - 9s 210ms/step - loss: 0.0567 - accuracy: 0.9765 - val_loss: 0.4833 - val_accuracy: 0.8419
Epoch 7/30
43/43 [==============================] - 9s 218ms/step - loss: 0.0312 - accuracy: 0.9897 - val_loss: 1.4513 - val_accuracy: 0.8034
Epoch 8/30
43/43 [==============================] - 9s 213ms/step - loss: 0.0820 - accuracy: 0.9707 - val_loss: 0.5821 - val_accuracy: 0.8248
Epoch 9/30
43/43 [==============================] - 9s 222ms/step - loss: 0.0513 - accuracy: 0.9897 - val_loss: 0.8516 - val_accuracy: 0.8462
Epoch 10/30
43/43 [==============================] - 11s 246ms/step - loss: 0.0442 - accuracy: 0.9853 - val_loss: 0.7927 - val_accuracy: 0.8397
Epoch 11/30
43/43 [==============================] - 10s 222ms/step - loss: 0.0356 - accuracy: 0.9897 - val_loss: 0.7730 - val_accuracy: 0.8141
Epoch 12/30
43/43 [==============================] - 10s 232ms/step - loss: 0.0309 - accuracy: 0.9824 - val_loss: 0.9528 - val_accuracy: 0.8226
Epoch 13/30
43/43 [==============================] - 9s 220ms/step - loss: 0.0424 - accuracy: 0.9839 - val_loss: 1.2109 - val_accuracy: 0.8013
Epoch 14/30
43/43 [==============================] - 10s 228ms/step - loss: 0.0645 - accuracy: 0.9824 - val_loss: 0.5308 - val_accuracy: 0.8547
Epoch 15/30
43/43 [==============================] - 11s 259ms/step - loss: 0.0293 - accuracy: 0.9927 - val_loss: 0.9271 - val_accuracy: 0.8333
Epoch 16/30
43/43 [==============================] - 9s 217ms/step - loss: 0.0430 - accuracy: 0.9795 - val_loss: 0.6687 - val_accuracy: 0.8483
I have tried many different model architectures, changing number of layers, kernel size etc... I can't seem to figure out what is going wrong.
There are many possible reasons.
For starters, depending on your categories, you might want to consider using transfer learning to speed up your training process.
Your architecture looks reasonable and the training and validation loss seems right as well (overfitting is occurring).
Given that you've stated that you could have 3 categories and am currently only using 2, might there be a different distribution between your training set and your test set? That might be causing the model to be unable to generalise well.
For instance, your dataset contains of evenly distributed number of images of Cats, Dogs and Humans. You set 2 categories to train on and thus your model attempts to segment between humans and animals when it tries to validate, there is an uneven distribution in the training data causing the model to see insufficient training size of humans (33%)?
Related
First of all, I know that there is a similar thread here:
https://stats.stackexchange.com/questions/352036/what-should-i-do-when-my-neural-network-doesnt-learn
But unfortunately, it does not help. I probably have a bug inside my code which I cannot find. What I am trying to do is to classify some WAV files. But the model does not learn.
At first, I am collecting the files and saving them in an array.
Second, create new directories, one for train data and one for val data.
Next, I am reading the WAV files, creating spectrograms, and saving them all to the train directory.
Afterward, I am moving 20% of the data from the train directory to the val directory.
Note: While creating the spectrograms I am checking the length of the WAV. If it is too short (less than 2 sec), I am doubling it. Out of this spectrogram, I am cutting a random chunk and saving only this. As a result, all images do have the same height and width.
Then as the next step, I am loading the train and val images. And here I am also doing the normalization.
IMG_WIDTH=300
IMG_HEIGHT=300
IMG_DIM = (IMG_WIDTH, IMG_HEIGHT, 3)
train_files = glob.glob(DBMEL_PATH + "*",recursive=True)
train_imgs = [img_to_array(load_img(img, target_size=IMG_DIM)) for img in train_files]
train_imgs = np.array(train_imgs) / 255 # normalizing Data
train_labels = [fn.split('\\')[-1].split('.')[1].strip() for fn in train_files]
validation_files = glob.glob(DBMEL_VAL_PATH + "*",recursive=True)
validation_imgs = [img_to_array(load_img(img, target_size=IMG_DIM)) for img in validation_files]
validation_imgs = np.array(validation_imgs) / 255 # normalizing Data
validation_labels = [fn.split('\\')[-1].split('.')[1].strip() for fn in validation_files]
I have checked the variables and printing them. I guess this is working quite well. The arrays contain 80% and respectively 20% of the total data.
#Train dataset shape: (3756, 300, 300, 3)
#Validation dataset shape: (939, 300, 300, 3)
Next, I have also implemented a One-Hot-Encoder.
So far so good. In the next step I create empty DataGenerators, so without any data augmentation. When calling the DataGenerators, one time for train-data and one time for val-data, I'll pass the arrays for images (train_imgs, validation_imgs) and the one-hot-encoded-labels (train_labels_enc, validation_labels_enc).
Okay. Here now comes the tricky part.
First, create/load a pre-trained network
from tensorflow.keras.applications.resnet50 import ResNet50
from tensorflow.keras.models import Model
import tensorflow.keras
input_shape=(IMG_HEIGHT,IMG_WIDTH,3)
restnet = ResNet50(include_top=False, weights='imagenet', input_shape=(IMG_HEIGHT,IMG_WIDTH,3))
output = restnet.layers[-1].output
output = tensorflow.keras.layers.Flatten()(output)
restnet = Model(restnet.input, output)
for layer in restnet.layers:
layer.trainable = False
And now finally creating the model itself. While creating the model I am using the pre-trained network for transfer learning. I guess somewhere there must be a problem.
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout, InputLayer
from tensorflow.keras.models import Sequential
from tensorflow.keras import optimizers
model = Sequential()
model.add(restnet) # <-- transfer learning
model.add(Dense(512, activation='relu', input_dim=input_shape))# 512 (num_classes)
model.add(Dropout(0.3))
model.add(Dense(512, activation='relu'))
model.add(Dropout(0.3))
model.add(Dense(7, activation='softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
model.summary()
And the models run with this
history = model.fit_generator(train_generator,
steps_per_epoch=100,
epochs=100,
validation_data=val_generator,
validation_steps=10,
verbose=1
)
But even after 50 epochs the accuracy stalls at around 0.15
Epoch 1/100
100/100 [==============================] - 711s 7s/step - loss: 10.6419 - accuracy: 0.1530 - val_loss: 1.9416 - val_accuracy: 0.1467
Epoch 2/100
100/100 [==============================] - 733s 7s/step - loss: 1.9595 - accuracy: 0.1550 - val_loss: 1.9372 - val_accuracy: 0.1267
Epoch 3/100
100/100 [==============================] - 731s 7s/step - loss: 1.9940 - accuracy: 0.1444 - val_loss: 1.9388 - val_accuracy: 0.1400
Epoch 4/100
100/100 [==============================] - 735s 7s/step - loss: 1.9416 - accuracy: 0.1535 - val_loss: 1.9380 - val_accuracy: 0.1733
Epoch 5/100
100/100 [==============================] - 737s 7s/step - loss: 1.9394 - accuracy: 0.1656 - val_loss: 1.9345 - val_accuracy: 0.1533
Epoch 6/100
100/100 [==============================] - 741s 7s/step - loss: 1.9364 - accuracy: 0.1667 - val_loss: 1.9286 - val_accuracy: 0.1767
Epoch 7/100
100/100 [==============================] - 740s 7s/step - loss: 1.9389 - accuracy: 0.1523 - val_loss: 1.9305 - val_accuracy: 0.1400
Epoch 8/100
100/100 [==============================] - 737s 7s/step - loss: 1.9394 - accuracy: 0.1623 - val_loss: 1.9441 - val_accuracy: 0.1667
Epoch 9/100
100/100 [==============================] - 735s 7s/step - loss: 1.9391 - accuracy: 0.1582 - val_loss: 1.9458 - val_accuracy: 0.1333
Epoch 10/100
100/100 [==============================] - 734s 7s/step - loss: 1.9381 - accuracy: 0.1602 - val_loss: 1.9372 - val_accuracy: 0.1700
Epoch 11/100
100/100 [==============================] - 739s 7s/step - loss: 1.9392 - accuracy: 0.1623 - val_loss: 1.9302 - val_accuracy: 0.2167
Epoch 12/100
100/100 [==============================] - 741s 7s/step - loss: 1.9368 - accuracy: 0.1627 - val_loss: 1.9326 - val_accuracy: 0.1467
Epoch 13/100
100/100 [==============================] - 740s 7s/step - loss: 1.9381 - accuracy: 0.1513 - val_loss: 1.9312 - val_accuracy: 0.1733
Epoch 14/100
100/100 [==============================] - 736s 7s/step - loss: 1.9396 - accuracy: 0.1542 - val_loss: 1.9407 - val_accuracy: 0.1367
Epoch 15/100
100/100 [==============================] - 741s 7s/step - loss: 1.9393 - accuracy: 0.1597 - val_loss: 1.9336 - val_accuracy: 0.1333
Epoch 16/100
100/100 [==============================] - 737s 7s/step - loss: 1.9375 - accuracy: 0.1659 - val_loss: 1.9354 - val_accuracy: 0.1267
Epoch 17/100
100/100 [==============================] - 741s 7s/step - loss: 1.9422 - accuracy: 0.1487 - val_loss: 1.9307 - val_accuracy: 0.1567
Epoch 18/100
100/100 [==============================] - 738s 7s/step - loss: 1.9399 - accuracy: 0.1680 - val_loss: 1.9408 - val_accuracy: 0.1567
Epoch 19/100
100/100 [==============================] - 743s 7s/step - loss: 1.9405 - accuracy: 0.1610 - val_loss: 1.9335 - val_accuracy: 0.1533
Epoch 20/100
100/100 [==============================] - 738s 7s/step - loss: 1.9410 - accuracy: 0.1575 - val_loss: 1.9331 - val_accuracy: 0.1533
Epoch 21/100
100/100 [==============================] - 746s 7s/step - loss: 1.9395 - accuracy: 0.1639 - val_loss: 1.9344 - val_accuracy: 0.1733
Epoch 22/100
100/100 [==============================] - 746s 7s/step - loss: 1.9393 - accuracy: 0.1585 - val_loss: 1.9354 - val_accuracy: 0.1667
Epoch 23/100
100/100 [==============================] - 746s 7s/step - loss: 1.9398 - accuracy: 0.1599 - val_loss: 1.9352 - val_accuracy: 0.1500
Epoch 24/100
100/100 [==============================] - 746s 7s/step - loss: 1.9392 - accuracy: 0.1585 - val_loss: 1.9449 - val_accuracy: 0.1667
Epoch 25/100
100/100 [==============================] - 746s 7s/step - loss: 1.9399 - accuracy: 0.1495 - val_loss: 1.9352 - val_accuracy: 0.1600
Can anyone please help to find the problem?
I solved the problem on my own.
I exchanged this
model = Sequential()
model.add(restnet) # <-- transfer learning
model.add(Dense(512, activation='relu', input_dim=input_shape))# 512 (num_classes)
model.add(Dropout(0.3))
model.add(Dense(512, activation='relu'))
model.add(Dropout(0.3))
model.add(Dense(7, activation='softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
model.summary()
with this:
base_model = tf.keras.applications.MobileNetV2(input_shape = (224, 224, 3), include_top = False, weights = "imagenet")
model = Sequential()
model.add(base_model)
model.add(tf.keras.layers.GlobalAveragePooling2D())
model.add(Dropout(0.2))
model.add(Dense(number_classes, activation="softmax"))
model.compile(optimizer=tf.keras.optimizers.Adam(lr=0.00001),
loss="categorical_crossentropy",
metrics=['accuracy'])
model.summary()
And I found out one more thing. In contrary to some tutorials, using data augmentation is not useful when working with spectrograms.
Without data augmentation I got 0.99 on train-accuracy and 0.72 on val-accuracy. But with data augmentation I got only 0.75 on train-accuracy and 0.16 on val-accuracy.
I'm trying to figure out why model's Loss value is always 0.0, so the accuracy seems to be constant as well (which is incorrect in my case, afaik).
Code snippet:
model = Sequential()
model.add(Embedding(vocab_size, glove_vectors.vector_size, weights=[embedding_matrix], input_length=X.shape[1]))
model.add(Dropout(0.5))
model.add(LSTM(100, dropout=0.2, recurrent_dropout=0.2))
model.add(Dense(1, activation="sigmoid"))
model.compile(loss='categorical_crossentropy', optimizer="adam", metrics=["accuracy"])
model.summary()
EPOCHS = 20
train_data, test_data, train_labels, test_labels = train_test_split(X, Y, test_size=0.20, random_state = 42)
print(train_data.shape, train_labels.shape)
print(test_data.shape, test_labels.shape)
val_data = (test_data, test_labels)
history = model.fit(train_data, train_labels, validation_data=val_data, epochs=EPOCHS)
score = model.evaluate(test_data, test_labels)
Output:
Epoch 1/20
25/25 [==============================] - 4s 69ms/step - loss: 0.0000e+00 - accuracy: 0.5241 - val_loss: 0.0000e+00 - val_accuracy: 0.4650
Epoch 2/20
25/25 [==============================] - 1s 55ms/step - loss: 0.0000e+00 - accuracy: 0.4927 - val_loss: 0.0000e+00 - val_accuracy: 0.4650
Epoch 3/20
25/25 [==============================] - 1s 55ms/step - loss: 0.0000e+00 - accuracy: 0.5110 - val_loss: 0.0000e+00 - val_accuracy: 0.4650
Epoch 4/20
25/25 [==============================] - 1s 56ms/step - loss: 0.0000e+00 - accuracy: 0.5074 - val_loss: 0.0000e+00 - val_accuracy: 0.4650
Epoch 5/20
25/25 [==============================] - 1s 55ms/step - loss: 0.0000e+00 - accuracy: 0.5363 - val_loss: 0.0000e+00 - val_accuracy: 0.4650
Epoch 6/20
25/25 [==============================] - 1s 53ms/step - loss: 0.0000e+00 - accuracy: 0.5042 - val_loss: 0.0000e+00 - val_accuracy: 0.4650
Epoch 7/20
25/25 [==============================] - 1s 54ms/step - loss: 0.0000e+00 - accuracy: 0.4904 - val_loss: 0.0000e+00 - val_accuracy: 0.4650
In binary classification there will be 1 node in the output layer even though we will be predicting between two classes. In order to get the output in a probability format between 0 and 1 we will use the sigmoid function.
Hence binary_crossentropy is the correct loss function in your case
model.compile(loss='binary_crossentropy', optimizer="adam", metrics=["accuracy"])
I'm new to Keras and I'm using it to build a normal Neural Network to classify number MNIST dataset.
Beforehand I have already split the data into 3 parts: 55000 to train, 5000 to evaluate and 10000 to test, and I have scaled the pixel density down (by dividing it by 255.0)
My model looks like this:
model = keras.models.Sequential()
model.add(keras.layers.Flatten(input_shape=[28,28]))
model.add(keras.layers.Dense(100, activation='relu'))
model.add(keras.layers.Dense(10, activation='softmax'))
And here is the compile:
model.compile(loss='sparse_categorical_crossentropy',
optimizer = 'Adam',
metrics=['accuracy'])
I train the model:
his = model.fit(xTrain, yTrain, epochs = 20, validation_data=(xValid, yValid))
At first the val_loss decreases, then it increases although the accuracy is increasing.
Train on 55000 samples, validate on 5000 samples
Epoch 1/20
55000/55000 [==============================] - 5s 91us/sample - loss: 0.2822 - accuracy: 0.9199 - val_loss: 0.1471 - val_accuracy: 0.9588
Epoch 2/20
55000/55000 [==============================] - 5s 82us/sample - loss: 0.1274 - accuracy: 0.9626 - val_loss: 0.1011 - val_accuracy: 0.9710
Epoch 3/20
55000/55000 [==============================] - 5s 83us/sample - loss: 0.0899 - accuracy: 0.9734 - val_loss: 0.0939 - val_accuracy: 0.9742
Epoch 4/20
55000/55000 [==============================] - 5s 84us/sample - loss: 0.0674 - accuracy: 0.9796 - val_loss: 0.0760 - val_accuracy: 0.9770
Epoch 5/20
55000/55000 [==============================] - 5s 94us/sample - loss: 0.0541 - accuracy: 0.9836 - val_loss: 0.0842 - val_accuracy: 0.9742
Epoch 15/20
55000/55000 [==============================] - 4s 82us/sample - loss: 0.0103 - accuracy: 0.9967 - val_loss: 0.0963 - val_accuracy: 0.9788
Epoch 16/20
55000/55000 [==============================] - 5s 84us/sample - loss: 0.0092 - accuracy: 0.9973 - val_loss: 0.0956 - val_accuracy: 0.9774
Epoch 17/20
55000/55000 [==============================] - 5s 82us/sample - loss: 0.0081 - accuracy: 0.9977 - val_loss: 0.0977 - val_accuracy: 0.9770
Epoch 18/20
55000/55000 [==============================] - 5s 85us/sample - loss: 0.0076 - accuracy: 0.9977 - val_loss: 0.1057 - val_accuracy: 0.9760
Epoch 19/20
55000/55000 [==============================] - 5s 83us/sample - loss: 0.0063 - accuracy: 0.9980 - val_loss: 0.1108 - val_accuracy: 0.9774
Epoch 20/20
55000/55000 [==============================] - 5s 85us/sample - loss: 0.0066 - accuracy: 0.9980 - val_loss: 0.1056 - val_accuracy: 0.9768
And when I evaluate the loss is too high:
model.evaluate(xTest, yTest)
Result:
10000/10000 [==============================] - 0s 41us/sample - loss: 25.7150 - accuracy: 0.9740
[25.714989705941953, 0.974]
Is this ok, or is it a sign of overfitting? Should I do something to improve it? Thanks in advance.
Usually, it is not Ok. You want the loss rate to be as small as possible. Your result is typical for overfitting. Your Network 'knows' its training data, but isn't capable of analysing new Images. You may want to add some layers. Maybe Convolutional Layers, Dropout Layer... another idea would be to augment your training images. The ImageDataGenerator-Class provided by Keras might help you out here
Another thing to look at could be your hyperparameters. Why do you use 100 nodes in the first dense layer? maybe something like 784 (28*28) seems more interesting if you want to start with a dense layer. I would suggest some combination of Convolutional-Dropout-Dense. Then your dense -layer maybe doesn't need that many nodes...
I've spend the last 2 weeks struggling with my NN. The aim is to predict trip durations of taxi courses based on several
numerical variables (latitudes and longitudes)
categorical variables (numerically encoded) (hour of the day, day of the week, etc)
Here is the simplest version
X_train = trainData.as_matrix(columns=["fareDistance","hour","day","pickup_longitude","pickup_latitude","dropoff_longitude","dropoff_latitude"])
Y_train = np.array(trainData["trip_duration"])
model = Sequential()
model.add(Dense(32, input_dim=7, activation='linear'))
model.add(Dense(12, activation='linear'))
model.add(Dense(1, activation='linear'))
model.compile(loss='mean_absolute_percentage_error', optimizer='adagrad', metrics=['accuracy'])
model.summary()
model.fit(X_train, Y_train, epochs=10, validation_split=0.2)
I also tried to merge two different models for numerical variables on one hand and categorical on the other but it didn't change a thing. Depending on the combinations of Loss and optimization function either the loss and accuracy remain quite the same (acc. 0.0016) or I don't even have non null acc.
A friend of mine replicated the NN in pure TensorFlow and got the same kind of results
Train on 233383 samples, validate on 58346 samples
Epoch 1/20 233383/233383 [==============================] - 15s - loss: 45.9550 - acc: 0.0016 - val_loss: 46.2514 - val_acc: 0.0014
Epoch 2/20 233383/233383 [==============================] - 15s - loss: 45.8675 - acc: 0.0014 - val_loss: 46.2675 - val_acc: 0.0015
Epoch 3/20 233383/233383 [==============================] - 15s - loss: 45.8465 - acc: 0.0015 - val_loss: 46.2131 - val_acc: 0.0013
Epoch 4/20 233383/233383 [==============================] - 15s - loss: 45.8283 - acc: 0.0014 - val_loss: 46.2478 - val_acc: 0.0016
Epoch 5/20 233383/233383 [==============================] - 15s - loss: 45.8214 - acc: 0.0015 - val_loss: 46.2043 - val_acc: 0.0013
Epoch 6/20 233383/233383 [==============================] - 14s - loss: 45.8122 - acc: 0.0014 - val_loss: 46.2526 - val_acc: 0.0014
Epoch 7/20 233383/233383 [==============================] - 12s - loss: 45.7990 - acc: 0.0015 - val_loss: 46.1821 - val_acc: 0.0014
Epoch 8/20 233383/233383 [==============================] - 12s - loss: 45.7964 - acc: 0.0016 - val_loss: 46.1761 - val_acc: 0.0013
Epoch 9/20 233383/233383 [==============================] - 11s - loss: 45.7898 - acc: 0.0015 - val_loss: 46.1804 - val_acc: 0.0016
Am I missing something -- like something big, obvious -- which would explain why any attempt to change activation, loss or optimization function ends up doing the same?
Thanks in advance
D.
try this:
X_train = trainData.as_matrix(columns=["fareDistance","hour","day","pickup_longitude","pickup_latitude","dropoff_longitude","dropoff_latitude"])
Y_train = np.array(trainData["trip_duration"])
model = Sequential()
model.add(Dense(32, input_dim=7, activation='elu'))
model.add(Dense(12, activation='elu'))
model.add(Dense(1, kernel_initializer='normal'))
model.compile(loss='mean_absolute_percentage_error', optimizer='rmsprop')
model.summary()
model.fit(X_train, Y_train, epochs=10, validation_split=0.2)
you can also try the adam optimizer.
model.compile(loss='mean_absolute_percentage_error', optimizer='adam')
Update:
If the code above didn't help you it means your input data either not normalized or very dirty.
I have the following neural network, written in Keras using Tensorflow as the backend, which I'm running on Python 3.5 (Anaconda) on Windows 10:
model = Sequential()
model.add(Dense(100, input_dim=283, init='normal', activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(150, init='normal', activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(200, init='normal', activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(200, init='normal', activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(200, init='normal', activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(4, init='normal', activation='sigmoid'))
sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])
I'm training on my GPU. During training (10000 epochs), the accuracy of the naive network steadily increases from 0.25 to somewhere between 0.7 and 0.9, before suddenly dropping and sticking at 0.25:
Epoch 1/10000
6120/6120 [==============================] - 1s - loss: 1.5329 - acc: 0.2665
Epoch 2/10000
6120/6120 [==============================] - 1s - loss: 1.2985 - acc: 0.3784
Epoch 3/10000
6120/6120 [==============================] - 1s - loss: 1.2259 - acc: 0.4891
Epoch 4/10000
6120/6120 [==============================] - 1s - loss: 1.1867 - acc: 0.5208
Epoch 5/10000
6120/6120 [==============================] - 1s - loss: 1.1494 - acc: 0.5199
Epoch 6/10000
6120/6120 [==============================] - 1s - loss: 1.1042 - acc: 0.4953
Epoch 7/10000
6120/6120 [==============================] - 1s - loss: 1.0491 - acc: 0.4982
Epoch 8/10000
6120/6120 [==============================] - 1s - loss: 1.0066 - acc: 0.5065
Epoch 9/10000
6120/6120 [==============================] - 1s - loss: 0.9749 - acc: 0.5338
Epoch 10/10000
6120/6120 [==============================] - 1s - loss: 0.9456 - acc: 0.5696
Epoch 11/10000
6120/6120 [==============================] - 1s - loss: 0.9252 - acc: 0.5995
Epoch 12/10000
6120/6120 [==============================] - 1s - loss: 0.9111 - acc: 0.6106
Epoch 13/10000
6120/6120 [==============================] - 1s - loss: 0.8772 - acc: 0.6160
Epoch 14/10000
6120/6120 [==============================] - 1s - loss: 0.8517 - acc: 0.6245
Epoch 15/10000
6120/6120 [==============================] - 1s - loss: 0.8170 - acc: 0.6345
Epoch 16/10000
6120/6120 [==============================] - 1s - loss: 0.7850 - acc: 0.6428
Epoch 17/10000
6120/6120 [==============================] - 1s - loss: 0.7633 - acc: 0.6580
Epoch 18/10000
6120/6120 [==============================] - 4s - loss: 0.7375 - acc: 0.6717
Epoch 19/10000
6120/6120 [==============================] - 1s - loss: 0.7058 - acc: 0.6850
Epoch 20/10000
6120/6120 [==============================] - 1s - loss: 0.6787 - acc: 0.7018
Epoch 21/10000
6120/6120 [==============================] - 1s - loss: 0.6557 - acc: 0.7093
Epoch 22/10000
6120/6120 [==============================] - 1s - loss: 0.6304 - acc: 0.7208
Epoch 23/10000
6120/6120 [==============================] - 1s - loss: 0.6052 - acc: 0.7270
Epoch 24/10000
6120/6120 [==============================] - 1s - loss: 0.5848 - acc: 0.7371
Epoch 25/10000
6120/6120 [==============================] - 1s - loss: 0.5564 - acc: 0.7536
Epoch 26/10000
6120/6120 [==============================] - 1s - loss: 0.1787 - acc: 0.4163
Epoch 27/10000
6120/6120 [==============================] - 1s - loss: 1.1921e-07 - acc: 0.2500
Epoch 28/10000
6120/6120 [==============================] - 1s - loss: 1.1921e-07 - acc: 0.2500
Epoch 29/10000
6120/6120 [==============================] - 1s - loss: 1.1921e-07 - acc: 0.2500
Epoch 30/10000
6120/6120 [==============================] - 2s - loss: 1.1921e-07 - acc: 0.2500
Epoch 31/10000
6120/6120 [==============================] - 1s - loss: 1.1921e-07 - acc: 0.2500
Epoch 32/10000
6120/6120 [==============================] - 1s - loss: 1.1921e-07 - acc: 0.2500 ...
I'm guessing that this is due to the optimiser falling into a local minimum where it assigns all data to one category. How can I inhibit it from doing this?
Things I've tried (but didn't seem to stop this from happening):
Using a different optimiser (adam)
Ensuring that the training data included an equal number of examples from each category
Increasing the volume of training data (currently at 6000)
Varying the number of categories between 2 to 5
Increasing the number of hidden layers in the network from 1 to 5
Changing the width of the layers (from 50 to 500)
None of these helped. Any other ideas why this is happening and/or how to inhibit it? Could it be a bug in Keras? Many thanks in advance for any suggestions.
Edit:
The problem appears to have been solved by changing the final activation to softmax (from sigmoid) and adding maxnorm(3) regularization to the final two hidden layers:
model = Sequential()
model.add(Dense(100, input_dim=npoints, init='normal', activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(150, init='normal', activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(200, init='normal', activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(200, init='normal', activation='relu', W_constraint=maxnorm(3)))
model.add(Dropout(0.2))
model.add(Dense(200, init='normal', activation='relu', W_constraint=maxnorm(3)))
model.add(Dropout(0.2))
sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
model.add(Dense(ncat, init='normal', activation='softmax'))
model.compile(loss='mean_squared_error', optimizer=sgd, metrics=['accuracy'])
Many thanks for the suggestions.
The problem lied in sigmoid function as an activation in a last layer. In this case the output of your final layer cannot be interpreted as a probability distribution of an example given belonging to a single class. The output from this layer usually doesn't even sum up to 1. In this case the optimization may lead to unexpected behaviour. In my opinion adding a maxnorm constrain is not necessary but I strongly advise you to use a categorical_crossentropy instead of mse loss as it's proven that this function works better for this optimization case.