I'm trying to figure out how to decrease the error in my LSTM. It's an odd use-case because rather than classifying, we are taking in short lists (up to 32 elements long) and outputting a series of real numbers, ranging from -1 to 1 - representing angles. Essentially, we want to reconstruct short protein loops from amino acid inputs.
In the past we had redundant data in our datasets, so the accuracy reported was incorrect. Since removing the redundant data our validation accuracy has gotten much worse, which suggests our network had learned to memorise the most frequent examples.
Our dataset is 10,000 items, split 70/20/10 between train, validation and test. We use a bi-directional, LSTM as follows:
x = tf.cast(tf_train_dataset, dtype=tf.float32)
output_size = FLAGS.max_cdr_length * 4
dmask = tf.placeholder(tf.float32, [None, output_size], name="dmask")
keep_prob = tf.placeholder(tf.float32, name="keepprob")
sizes = [FLAGS.lstm_size,int(math.floor(FLAGS.lstm_size/2)),int(math.floor(FLAGS.lstm_size/ 4))]
single_rnn_cell_fw = tf.contrib.rnn.MultiRNNCell( [lstm_cell(sizes[i], keep_prob, "cell_fw" + str(i)) for i in range(len(sizes))])
single_rnn_cell_bw = tf.contrib.rnn.MultiRNNCell( [lstm_cell(sizes[i], keep_prob, "cell_bw" + str(i)) for i in range(len(sizes))])
length = create_length(x)
initial_state = single_rnn_cell_fw.zero_state(FLAGS.batch_size, dtype=tf.float32)
initial_state = single_rnn_cell_bw.zero_state(FLAGS.batch_size, dtype=tf.float32)
outputs, states = tf.nn.bidirectional_dynamic_rnn(cell_fw=single_rnn_cell_fw, cell_bw=single_rnn_cell_bw, inputs=x, dtype=tf.float32, sequence_length = length)
output_fw, output_bw = outputs
states_fw, states_bw = states
output_fw = last_relevant(FLAGS, output_fw, length, "last_fw")
output_bw = last_relevant(FLAGS, output_bw, length, "last_bw")
output = tf.concat((output_fw, output_bw), axis=1, name='bidirectional_concat_outputs')
test = tf.placeholder(tf.float32, [None, output_size], name="train_test")
W_o = weight_variable([sizes[-1]*2, output_size], "weight_output")
b_o = bias_variable([output_size],"bias_output")
y_conv = tf.tanh( ( tf.matmul(output, W_o)) * dmask, name="output")
Essentially, we use 3 layers of LSTM, with 256, 128 and 64 units each. We take the last step of both the Forward and Backward passes and concatenate them together. These feed into a final, fully connected layer that presents the data in the way we need it. We use a mask to set these steps we don't need to zero.
Our cost function uses a mask again, and takes the mean of the squared difference. We build the mask from the test data. Values to ignore are set to -3.0.
def cost(goutput, gtest, gweights, FLAGS):
mask = tf.sign(tf.add(gtest,3.0))
basic_error = tf.square(gtest-goutput) * mask
basic_error = tf.reduce_sum(basic_error)
basic_error /= tf.reduce_sum(mask)
return basic_error
To train the net I've used a variety of optimizers. The lowest scores have been obtained with the AdamOptimizer. The others, such as Adagrad, Adadelta, RMSProp tend to flatline around 0.3/0.4 error which is not particularly great.
Our learning rate is 0.004, batch size of 200. We use a 0.5 probability dropout layer.
I've tried adding more layers, changing learning rates, batch sizes, even the representation of the data. I've attempted batch regularisation, L1 and L2 weight regularisation (though perhaps incorrectly) and I've even considered switching to a convnet approach instead.
Nothing seems to make any difference. What has seemed to work is changing the optimizer. Adam seems noisier as it improves, but it does get closer than the other optimizers.
We need to get down to a value much closer to 0.05 or 0.01. Sometimes the training error touches 0.09 but the validation doesn't follow. I've run this network for about 500 epochs so far (about 8 hours) and it tends to settle around 0.2 validation error.
I'm not quite sure what to attempt next. Decayed learning rate might help but I suspect there is something more fundamental I need to do. It could be something as simple as a bug in the code - I need to double check the masking,
Related
I'm making a model for predicting the age of people by analyzing their face. I'm using this pretrained model, and maked a custom loss function and a custom metrics. So I obtain discrete result but I want to improve it. In particular, I noticed that after some epochs the model begin to overfitt on the training set then the val_loss increases. How can I avoid this? I'm already using Dropout, but this doesn't seem to be enough.
I think maybe I should use l1 and l2 but I don't know how.
def resnet_model():
model = VGGFace(model = 'resnet50')#model :{resnet50, vgg16, senet50}
xl = model.get_layer('avg_pool').output
x = keras.layers.Flatten(name='flatten')(xl)
x = keras.layers.Dense(4096, activation='relu')(x)
x = keras.layers.Dropout(0.5)(x)
x = keras.layers.Dense(4096, activation='relu')(x)
x = keras.layers.Dropout(0.5)(x)
x = keras.layers.Dense(11, activation='softmax', name='predictions')(x)
model = keras.engine.Model(model.input, outputs = x)
return model
model = resnet_model()
initial_learning_rate = 0.0003
epochs = 20; batch_size = 110
num_steps = train_x.shape[0]//batch_size
learning_rate_fn = tf.keras.optimizers.schedules.PiecewiseConstantDecay(
[3*num_steps, 10*num_steps, 16*num_steps, 25*num_steps],
[1e-4, 1e-5, 1e-6, 1e-7, 5e-7]
)
optimizer = tf.keras.optimizers.Adam(learning_rate=learning_rate_fn)
model.compile(loss=custom_loss, optimizer=optimizer, metrics=['accuracy', one_off_accuracy])
model.fit(train_x, train_y, epochs=epochs, batch_size=batch_size, validation_data=(test_x, test_y))
This is an example of result:
There are many regularization methods to help you avoid overfitting your model:
Dropouts:
Randomly disables neurons during the training, in order to force other neurons to be trained as well.
L1/L2 penalties:
Penalizes weights that change dramatically. This tries to ensure that all parameters will be equally taken into consideration when classifying an input.
Random Gaussian Noise at the inputs:
Adds random gaussian noise at the inputs: x = x + r where r is a random normal value from range [-1, 1]. This will confuse your model and prevent it from overfitting into your dataset, because in every epoch, each input will be different.
Label Smoothing:
Instead of saying that a target is 0 or 1, You can smooth those values (e.g. 0.1 & 0.9).
Early Stopping:
This is a quite common technique for avoiding training your model too much. If you notice that your model's loss is decreasing along with the validation's accuracy, then this is a good sign to stop the training, as your model begins to overfit.
K-Fold Cross-Validation:
This is a very strong technique, which ensures that your model is not fed all the time with the same inputs and is not overfitting.
Data Augmentations:
By rotating/shifting/zooming/flipping/padding etc. an image you make sure that your model is forced to train better its parameters and not overfit to the existing dataset.
I am quite sure there are also more techniques to avoid overfitting. This repository contains many examples of how the above techniques are deployed in a dataset:
https://github.com/kochlisGit/Tensorflow-State-of-the-Art-Neural-Networks
You can try incorporate image augmentation in your training, which increases the "sample size" of your data as well as the "diversity" as #Suraj S Jain mentioned. The official tutorial is here: https://www.tensorflow.org/tutorials/images/data_augmentation
I trained a simple fully connected network on CIFAR-10 dataset:
import torch
import torch.nn as nn
import torch.nn.functional as F
import torchvision
import torchvision.transforms as transforms
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.fc1 = nn.Linear(3*32*32, 300, bias=False)
self.fc2 = nn.Linear(300, 10, bias=False)
def forward(self, x):
x = x.reshape(250, -1)
self.x2 = F.relu(self.fc1(x))
x = self.fc2(self.x2)
return x
def train():
# The output of torchvision datasets are PILImage images of range [0, 1].
transform = transforms.Compose([transforms.ToTensor()])
trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=250, shuffle=True, num_workers=4)
testset = torchvision.datasets.CIFAR10(root='./data', train=False, download=True, transform=transform)
testloader = torch.utils.data.DataLoader(testset, batch_size=args.bs, shuffle=False, num_workers=4)
net = Net()
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(net.parameters(), lr=0.02, momentum=0.9, weight_decay=0.0001)
for epoch in range(20):
correct = 0
total = 0
for data in trainloader:
inputs, labels = data
outputs = net(inputs)
loss = criterion(outputs, labels)
optimizer.zero_grad()
loss.backward()
optimizer.step()
_, predicted = torch.max(outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
acc = 100. * correct / total
This network gets to ~50% test accuracy with the parameters specified, after 20 epochs.
Note that I didn't do any whitening of the inputs (no per channel mean subtraction)
Next I scaled up the model inputs by 255, by replacing outputs = net(inputs) with outputs = net(inputs*255). After this change, the network no longer converges. I looked at the gradients and they seem to grow explosively after just a few iterations, leading to all model outputs being zero. I'd like to understand why this is happening.
Also, I tried scaling down the learning rate by 255. This helps, but the network only gets to ~43% accuracy. Again, I don't understand why this helps, and more importantly why the accuracy is still degraded compared to the original settings.
EDIT: forgot to mention that I don't use biases in this network.
EDIT2: I can recover the original accuracy if I scale down the initial weights in both layers by 255 (in addition to scaling down the learning rate). I also tried to scale down the initial weights only in the first layer, but the network had trouble learning (even when I did scale down the learning rate in both layers). Then I tried scaling down the learning rate only in the first layer - this also didn't help. Finally I tried reducing learning rate in both layer even more (by 255*255) and this suddenly worked. This does not make sense to me - scaling down the initial weights by the same factor the inputs have been scaled up should have completely eliminated any difference from the original network, the input to the second layer is identical. At that point the learning rate should be scaled down in the first layer only, but in practice both layers need significantly lower learning rate...
Scaling up the inputs will lead to exploding gradients because of a few observations:
The learning rate is common to all the weights in a given update step.
Hence, the same scaling factor (ie: the learning rate) is applied to a given weight's cost derivative regardless of it's magnitude, so large and small weights get updated by the same scale.
When the loss landscape is highly erratic, this leads to exploding gradients.(like a snowball effect, one overshot update - in say, the axis of one particular weight - causes another in the opposite direction in the next update which overshoots again and so on..)
The range of values of the pixels are 0 to 255, hence scaling the data by 255 will ensure all inputs are between 0 and 1 and hence more smooth convergence as all the gradients will be uniform with respect to the learning rate. But here you scaled the learning rate which adjusts some of the problems mentioned above but is not as effective as scaling the data itself. This reduces the learning rate hence making convergence time longer, that might be the reason why it reaches 43% at 20 epochs, maybe it needs more epochs..
Also:
CIFAR-10 is a significant step up from something like the MNIST dataset, hence, fully connected neural networks do not have the representation power needed to accurately predict these images. CNNs are the way to go for any image classification task beyond MNIST. ~50% accuracy is the max you can get with a fully connected neural network unfortunately.
Maybe decrease the learning rate by 1/255 ... just a guess
I'm currently working on data with rare binary outcome, i.e. the response vector contains mostly 0 and only a few 1 (approximately 1.5% ones). I've got about 20 continuous explanatory variables. I tried to train models using GBM, Random Forests, TensorFlow with Keras backend.
I observed a special behavior of the models, regardless which method I used:
The accuracy is high (~98%) but the model predicts probabilities for class "0" for all outcomes as ~98.5% and for class "1" ~1,5%.
How can I prevent this behavior?
I'm using RStudio. For Example a TF model with Keras would be:
model <- keras_model_sequential()
model %>%
layer_dense(units = 256, activation = "relu", input_shape = c(20)) %>%
layer_dense(units = 256, activation = "relu") %>%
layer_dense(units = 2, activation = "sigmoid")
parallel_model <- multi_gpu_model(model, gpus=2)
parallel_model %>% compile(
optimizer = "adam",
loss = "binary_crossentropy",
metrics = "binary_accuracy")
histroy <- parallel_model %>% fit(
x_train, y_train,
batch_size = 64,
epochs = 100,
class_weight = list("0"=1,"1"=70),
verbose = 1,
validation_split = 0.2
)
But my observation is not limited to TF. This makes my question more general. I'm not asking for specific adjustments for the model above, rather I'd like to discuss at what point all outcomes are assigned the same probability.
I can guess, the issue is connected to the loss-function.
I know there is no way to use AUC as loss functions, since it's not differentiable. If I test models with AUC with unknown data, the result is not better than random guessing.
I don't mind answers with code in Python, since this isn't a problem about coding rather than general behavior and algorithms.
When your problem has unbalanced classes, I suggest using SMOTE (on the training data only!!! never use smote on your testing data!!!) before training the model.
For example:
from imblearn.over_sampling import SMOTE
X_trn_balanced, Y_trn_balanced = SMOTE(random_state=1, ratio=1).fit_sample(X_trn, Y_trn)
#next fit the model with the balanced data
model.fit(X_trn_balanced, Y_trn_balanced )
In my (not so big) experience with AUC problems and rare positives, I see models with one class (not two). It's either "result is positive (1)" or "result is negative (0)".
Metrics like accuracy are useless for these problems, you should use AUC based metrics with big batch sizes.
For these problems, it doesn't matter whether the outcome probabilities are too little, as long as there is a difference between them. (Forests, GBM, etc. will indeed output these little values, but this is not a problem)
For neural networks, you can try to use class weights to increase the output probabilities. But notice that if you split the result in two separate classes (considering only one class should be positive), it doesn't matter if you use weights, because:
For the first class, low weights: predict all ones is good
For the second class, high weights: predict all zeros is good (weighted to very good)
So, as an initial solution, you can:
Use a 'softmax' activation (to guarantee your model will have only one correct output) and a 'categorical_crossentropy' loss.
(Or, preferrably) Use a model with only one class and keep 'sigmoid' with 'binary_crossentropy'.
I always work with the preferrable option above. In this case, if you use batch sizes that are big enough to contain one or two positive examples (batch size around 100 for you), weights may even be discarded. If the batch sizes are too little and many batches don't contain positive results, you may have too many weight updates towards plain zeros, which is bad.
You may also resample your data and, for instance, multiply by 10 the number of positive examples, so your batches contain more positives and training becomes easier.
Example of AUC metric to determine when training should end:
#in python - considering outputs with only one class
def aucMetric(true, pred):
true= K.flatten(true)
pred = K.flatten(pred)
totalCount = K.shape(true)[0]
values, indices = tf.nn.top_k(pred, k = totalCount)
sortedTrue = K.gather(true, indices)
tpCurve = K.cumsum(sortedTrue)
negatives = 1 - sortedTrue
auc = K.sum(tpCurve * negatives)
totalCount = K.cast(totalCount, K.floatx())
positiveCount = K.sum(true)
negativeCount = totalCount - positiveCount
totalArea = positiveCount * negativeCount
return auc / totalArea
I used tensorflow to implement a simple RNN model to learn possible trends of time series data and predict future values. However, the model always produces same values after training. Actually, the best model it got is:
y = b.
The RNN structure is:
InputLayer -> BasicRNNCell -> Dense -> OutputLayer
RNN code:
def RNN(n_timesteps, n_input, n_output, n_units):
tf.reset_default_graph()
X = tf.placeholder(dtype=tf.float32, shape=[None, n_timesteps, n_input])
cells = [tf.contrib.rnn.BasicRNNCell(num_units=n_units)]
stacked_rnn = tf.contrib.rnn.MultiRNNCell(cells)
stacked_output, states = tf.nn.dynamic_rnn(stacked_rnn, X, dtype=tf.float32)
stacked_output = tf.layers.dense(stacked_output, n_output)
return X, stacked_output
while in training, n_timesteps=1, n_input=1, n_output=1, n_units=2, learning_rate=0.0000001. And loss is calculated by mean squared error.
Input is a sequence of data in continuous days. Output is the data after the days of input.
(Maybe these are not good settings. But no matter how I change them, the results are almost same. So I just set these to help show them later.)
And I found out this is because weights and bias of BasicRNNCell are not trained. They keep same from beginning. And only the weights and bias of Dense keep changing. So in training, I got a prediction like these:
In the beginning:
loss: 1433683500.0
rnn/multi_rnn_cell/cell_0/cell0/kernel:0 [KEEP UNCHANGED]
rnn/multi_rnn_cell/cell_0/cell0/bias:0 [KEEP UNCHANGED]
dense/kernel:0 [CHANGING]
dense/bias:0 [CHANGING]
After a while:
loss: 175372340.0
rnn/multi_rnn_cell/cell_0/cell0/kernel:0 [KEEP UNCHANGED]
rnn/multi_rnn_cell/cell_0/cell0/bias:0 [KEEP UNCHANGED]
dense/kernel:0 [CHANGING]
dense/bias:0 [CHANGING]
The orange line indicates the true data, the blue line indicates results of my code. Through training, the blue line will keep going up until model gets a stable loss.
So I doubt whether I did a wrong implementation, so I generate a group of data with y = 10x + 5 for testing. This time, My model learns the correct results.
In the beginning:
In the end:
I have tried:
add more layers of both BasicRNNCell and Dense
increase rnn cell hidden num(n_units) to 128
decrease learning_rate to 1e-10
increase timesteps to 60
They all dont work.
So, my questions are:
Is it because my model is too simple? But I think the trend of my data is not so complicated to learn. At least something like y = ax + b will produce a smaller loss than y = b.
What may lead to these results?
Or how should I go on debugging?
And now, I double maybe BasicRNNCell is not fully realized, users should implement some functions of it? I have no experience with tensorflow before.
It seems your net is just not fit for that kind of data, or from another point of view, your data is badly scaled. When adding the 4 lines below after the split_data, I get some sort of learning behavior, similar to the one with the a*x+b case
data = read_data(work_dir, input_file)
plot_data(data)
input_data, output_data, n_batches = split_data(data, n_timesteps, n_input, n_output)
# scale input and output data
input_data = input_data-input_data[0]
input_data = input_data/np.max(input_data)*1000
output_data = output_data-output_data[0]
output_data = output_data/np.max(output_data)*1000
I am new to tensorflow and am learning the basics at the moment so please bear with me.
My problem concerns strange non-convergent behaviour of neural networks when presented with the supposedly simple task of finding a regression function for a small training set consisting only of m = 100 data points {(x_1, y_1), (x_2, y_2),...,(x_100, y_100)}, where x_i and y_i are real numbers.
I first constructed a function that automatically generates a computational graph corresponding to a classical fully connected feedforward neural network:
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
import math
def neural_network_constructor(arch_list = [1,3,3,1],
act_func = tf.nn.sigmoid,
w_initializer = tf.contrib.layers.xavier_initializer(),
b_initializer = tf.zeros_initializer(),
loss_function = tf.losses.mean_squared_error,
training_method = tf.train.GradientDescentOptimizer(0.5)):
n_input = arch_list[0]
n_output = arch_list[-1]
X = tf.placeholder(dtype = tf.float32, shape = [None, n_input])
layer = tf.contrib.layers.fully_connected(
inputs = X,
num_outputs = arch_list[1],
activation_fn = act_func,
weights_initializer = w_initializer,
biases_initializer = b_initializer)
for N in arch_list[2:-1]:
layer = tf.contrib.layers.fully_connected(
inputs = layer,
num_outputs = N,
activation_fn = act_func,
weights_initializer = w_initializer,
biases_initializer = b_initializer)
Phi = tf.contrib.layers.fully_connected(
inputs = layer,
num_outputs = n_output,
activation_fn = tf.identity,
weights_initializer = w_initializer,
biases_initializer = b_initializer)
Y = tf.placeholder(tf.float32, [None, n_output])
loss = loss_function(Y, Phi)
train_step = training_method.minimize(loss)
return [X, Phi, Y, train_step]
With the above default values for the arguments, this function would construct a computational graph corresponding to a neural network with 1 input neuron, 2 hidden layers with 3 neurons each and 1 output neuron. The activation function is per default the sigmoid function. X corresponds to the input tensor, Y to the labels of the training data and Phi to the feedforward output of the neural network. The operation train_step performs one gradient-descent step when executed in the session environment.
So far, so good. If I now test a particular neural network (constructed with this function and the exact default values for the arguments given above) by making it learn a simple regression function for artificial data extracted from a sinewave, strange things happen:
Before training, the network seems to be a flat line. After 100.000 training iterations, it manages to partially learn the function, but only the part which is closer to 0. After this, it becomes flat again. Further training does not decrease the loss function anymore.
This get even stranger, when I take the exact same data set, but shift all x-values by adding 500:
Here, the network completely refuses to learn. I cannot understand why this is happening. I have tried changing the architecture of the network and its learning rate, but have observed similar effects: the closer the x-values of the data cloud are to the origin, the easier the network can learn. After a certain distance to the origin, learning stops completely. Changing the activation function from sigmoid to ReLu has only made things worse; here, the network tends to just converge to the average, no matter what position the data cloud is in.
Is there something wrong with my implementation of the neural-network-constructor? Or does this have something do do with initialization values? I have tried to get a deeper understanding of this problem now for quite a while and would greatly appreciate some advice. What could be the cause of this? All thoughts on why this behaviour is occuring are very much welcome!
Thanks,
Joker