I trained a classification network using tensorFlow with batch normalization in every convolutional layer. When I predict on a balanced test set where every category included in it, the accuracy is normal. However, if I chose any one specific category from test set, the accuracy is low, even zero.
But when 3 categories included in test set, the accuracy became higher. As we all know, the weights was fixed when the model finished training. But I find the balance in test set have greatly influence on prediction accuracy.
I think if batch normalization has influence on this, so I remove all batch normalization and retrained the model again. This time, when I predict only one category picture, it became normal.
Could anyone know why? THANKS!
You're right. If your training set is unbalanced you compute and accumulate mean values (for every layer) that are skewed in favor of the majority class.
In fact, you're not "normalizing" but instead, you're making the unbalancing problem worse.
Use batch normalization when you have a balanced training set and you can be sure that your batches will contain a balanced number of samples. This gives you optimal results.
However, since you added in the comments that you're using tf.contrib.layers.conv2d(x, num_output, kernel_size, stride, padding, activation_fn, normal_fn=tf.contrib.layers.batch_norm)
I spotted the problem: normalizer_fn calls the function you pass (batch_norm). But it uses the defaults parameters. By default, is_training equals to True thus you're computing even during the test phase the mean and the variance over the batch. Just read carefully the documentation of tf.contrib.layers.conv2d and use normalizer_params to pass is_training=True when training and is_training=False when testing/validating.
Related
Hello everyone so I made this cnn model.
My data:
Train folder->30 classes->800 images each->24000 all together
Validation folder->30 classes->100 images each->3000 all together
Test folder->30 classes -> 100 images each -> 3000 all together
-I've applied data augmentation. ( on the train data)
-I got 5 conv layers with filters 32->64->128->128->128
each with maxpooling and batch normalization
-Added dropout 0.5 after flattening layers
Train part looks good. Validation part has a lot of 'jumps' though. Does it overfit?
Is there any way to fix this and make validation part more stable?
Note: I plann to increase epochs on my final model I'm just experimenting to see what works best since the model takes a lot of time in order to train. So for now I train with 20 epochs.
I've applied data augmentation (on the train data).
What does this mean? What kind of data did you add and how much? You might think I'm nitpicking, but if the distribution of the augmented data is different enough from the original data, then this will indeed cause your model to generalize poorly to the validation set.
Increasing your epochs isn't going to help here, your training loss is already decreasing reasonably. Training your model for longer is a good step if the validation loss is also decreasing nicely, but that's obviously not the case.
Some things I would personally try:
Try decreasing the learning rate.
Try training the model without the augmented data and see how the validation loss behaves.
Try splitting the augmented data so that it's also contained in the validation set and see how the model behaves.
Train part looks good. Validation part has a lot of 'jumps' though. Does it overfit?
the answer is yes. The so-called 'jumps' in the validation part may indicate that the model is not generalizing well to the validation data and therefore your model might be overfitting.
Is there any way to fix this and make validation part more stable?
To fix this you can use the following:
Increasing the size of your training set
Regularization techniques
Early stopping
Reduce the complexity of your model
Use different hyperparameters like learning rate
I have some data (reuters).
I combine the training and test data into one large data set.
I shuffle the entire data set.
I then split the data set into 50%. 50% for training and 50% for testing.
I then setup a sequential model (Embedding, GRU, BatchNormalization, Dense).
I compile with adam optimizer, and loss = sparse_categorical_crossentropy
I run fit on it, for 5 epochs, with validation_split = 0.2
I get a val_acc of 95%. I am very happy with this.
I then call evaluate with the test data, and get acc = 71%
When I repeat evaluate with the train data (instead of test data), just to see what happens, I get 95% acc (as I should).
I am trying to understand what is wrong with my acc score with the test data.
The data is shuffled between train and test each time, so it can not be related to the data.
I tried the checkpoint save/restore trick, that did not seem to help.
I have reduced epochs in case of overfitting, but this has not helped.
I am curious what is going on. The only thing I can think of is there is some sort of overfitting going on that is carrying over to test, but is NOT impacting val_acc.
(Side note: the 20% data saved for validation during fit should not cause an overfitting problem, correct?)
Any ideas what could be wrong with my approach?
Thank you.
Edit 1:
Okey, I might be onto something. I noticed something odd about the test data that is included with the reuters dataset, vs the training data.
In previous experiments, results were always poor evaluating against test set, vs an unused portion of the training data.
This time, I combined the training and test data sets into one set, and shuffled.
Then shuffled some more, and then generated pseudo data around it, and shuffled some more.
Then I peeled off 20% to use as test data.
This time it worked. I got 95% for training, validation and evaluation accuracy.
I tried searching for information on the test set to see if anyone came up with anything but found no results. So I'm going to presently chalk it up to test data that is significantly different from the training set.
Edit 2:
Nevermind edit 1. I think I was corrupting my test data with pseudo-generated version of training data, that was close enough to work.
The only conclusion I can draw is that there is a lot of overfitting going on during my training AND using validation data during training is misleading, as it is also being overfit.
(Why validation data is being used to help overfit, I do not know).
Overfitting occurs due to many reasons, to overcome this problem you can try different methods like:
L1/L2 regularization.
Dropout layers: By using dropout layers in the network, we ignore a subset of units of our network with a set probability. Using dropout, we can reduce interdependent learning among units, which may have led to overfitting.
Early stopping: Monitors the performance of the model for every epoch on a held-out validation set during the training, and terminates the training conditional on the validation performance.
Feature selection: Improves the machine learning process and increases the predictive power of machine learning algorithms by selecting the most important variables and eliminating unnecessary and irrelevant features.
For more details refer to this link.
I have used 100000 samples to train a general model in Keras and achieve good performance. Then, for a particular sample, I want to use the trained weights as initialization and continue to optimize the weights to further optimize the loss of the particular sample.
However, the problem occurred. First, I load the trained weight by the keras API easily, then, I evaluate the loss of the one particular sample, and the loss is close to the loss of the validation loss during the training of the model. I think it is normal. However, when I use the trained weight as the inital and further optimize the weight over the one sample by model.fit(), the loss is really strange. It is much higher than the evaluate result and gradually became normal after several epochs.
I think it is strange that, for the same one simple and loading the same model weight, why the model.fit() and model.evaluate() return different results. I used batch normalization layers in my model and I wonder that it may be the reason. The result of model.evaluate() seems normal, as it is close to what I seen in the validation set before.
So what cause the different between fit and evaluation? How can I solve it?
I think your core issue is that you are observing two different loss values during fit and evaluate. This has been extensively discussed here, here, here and here.
The fit() function loss includes contributions from:
Regularizers: L1/L2 regularization loss will be added during training, increasing the loss value
Batch norm variations: during batch norm, running mean and variance of the batch will be collected and then those statistics will be used to perform normalization irrespective of whether batch norm is set to trainable or not. See here for more discussion on that.
Multiple batches: Of course, the training loss will be averaged over multiple batches. So if you take average of first 100 batches and evaluate on the 100th batch only, the results will be different.
Whereas for evaluate, just do forward propagation and you get the loss value, nothing random here.
Bottomline is, you should not compare train and validation loss (or fit and evaluate loss). Those functions do different things. Look for other metrics to check if your model is training fine.
I have a FFNN with 2 hidden layers for a regression task that overfits almost immediately (epoch 2-5, depending on # hidden units). (ReLU, Adam, MSE, same # hidden units per layer, tf.keras)
32 neurons:
128 neurons:
I will be tuning the number of hidden units, but to limit the search space I would like to know what the upper and lower bounds should be.
Afaik it is better to have a too large network and try to regularize via L2-reg or dropout than to lower the network's capacity -- because a larger network will have more local minima, but the actual loss value will be better.
Is there any point in trying to regularize (via e.g. dropout) a network that overfits from the get-go?
If so I suppose I could increase both bounds. If not I would lower them.
model = Sequential()
model.add(Dense(n_neurons, 'relu'))
model.add(Dense(n_neurons, 'relu'))
model.add(Dense(1, 'linear'))
model.compile('adam', 'mse')
Hyperparameter tuning is generally the hardest step in ML, In general we try different values randomly and evalute the model and choose those set of values which give the best performance.
Getting back to your question, You have a high varience problem (Good in training, bad in testing).
There are eight things you can do in order
Make sure your test and training distribution are same.
Make sure you shuffle and then split the data into two sets (test and train)
A good train:test split will be 105:15K
Use a deeper network with Dropout/L2 regularization.
Increase your training set size.
Try Early Stopping
Change your loss function
Change the network architecture (Switch to ConvNets, LSTM etc).
Depending on your computation power and time you can set a bound to the number of hidden units and hidden layers you can have.
because a larger network will have more local minima.
Nope, this is not quite true, in reality as the number of input dimension increases the chance of getting stuck into a local minima decreases. So We usually ignore the problem of local minima. It is very rare. The derivatives across all the dimensions in the working space must be zero for a local/global minima. Hence, it is highly unlikely in a typical model.
One more thing, I noticed you are using linear unit for last layer. I suggest you to go for ReLu instead. In general we do not need negative values in regression. It will reduce test/train error
Take this :
In MSE 1/2 * (y_true - y_prediction)^2
because y_prediction can be nagative value. The whole MSE term may blow up to large values as y_prediction gets highly negative or highly positive.
Using a ReLu for last layer makes sure that y_prediction is positive. Hence low error will be expected.
Let me try to substantiate some of the ideas here, referenced from Ian Goodfellow et. al. Deep Learning book which is available for free online:
Chapter 7: Regularization The most important point is data, one can and should avoid regularization if they have large amounts of data that best approximate the distribution. In you case, it looks like there might be a significant discrepancy between training and test data. You need to ensure the data is consistent.
Section 7.4: Data-augmentation With regards to data, Goodfellow talks about data-augmentation and inducing regularization by injecting noise (most likely Gaussian) which mathematically has the same effect. This noise works well with regression tasks as you limit the model from latching onto a single feature to overfit.
Section 7.8: Early Stopping is useful if you just want a model with the best test error. But again this only works if your data allows the training to infer the test data. If there is an immediate increase in test error the training would stop immediately.
Section 7.12: Dropout Just applying dropout to a regression model doesn't necessarily help. In fact "when extremely few labeled training examples are available, dropout is less effective". For classification, dropout forces the model to not rely on single features, but in regression all inputs might be required to compute a value rather than classify.
Chapter 11: Practicals emphasises the use of base models to ensure that the training task is not trivial. If a simple linear regression can achieve similar behaviour than you don't even have a training problem to begin with.
Bottom line is you can't just play with the model and hope for the best. Check the data, understand what is required and then apply the corresponding techniques. For more details read the book, it's very good. Your starting point should be a simple regression model, 1 layer, very few neurons and see what happens. Then incrementally experiment.
I'm using TensorFlow for a multi-target regression problem. Specifically, in a convolutional network with pixel-wise labeling with the input being an image and the label being a "heat-map" where each pixel has a float value. More specifically, the ground truth labeling for each pixel is lower bounded by zero, and, while technically having no upper bound, usually gets no larger than 1e-2.
Without batch normalization, the network is able to give a reasonable heat-map prediction. With batch normalization, the network takes much long to get to reasonable loss value, and the best it does is making every pixel the average value. This is using the tf.contrib.layers conv2d and batch_norm methods, with the batch_norm being passed to the conv2d's normalization_fn (or not in the case of no batch normalization). I had briefly tried batch normalization on another (single value) regression network, and had trouble then as well (though, I hadn't tested that as extensively). Is there a problem using batch normalization on regression problems in general? Is there a common solution?
If not, what could be some causes batch normalization failing on such an application? I've attempted a variety of initializations, learning rates, etc. I would expect the final layer (which of course does not use batch normalization) could use weights to scale the output of the penultimate layer to the appropriate regression values. Failing that, I removed batch norm from that layer, but with no improvement. I've attempted a small classification problem using batch normalization and saw no problem there, so it seems reasonable that it could be due somehow to the nature of the regression problem, but I don't know how that could cause such a drastic difference. Is batch normalization known to have trouble on regression problems?
I believe your issue is in the labels. Batch norm will scale all input values between 0 and 1. If the labels are not scaled to a similar range the task will be more difficult. This is because it requires the NN to learn values of a different scale.
By removing the batch norm from the penultimate layer, the task may be improved slightly, but you are still requiring an NN layer to learn to downscale values of its input while subsequently normalizing back to the range 0 - 1 (opposite to your objective).
To solve this problem, apply a 0 - 1 scaler to the labels such that your upper bound is no longer 1e-2. During inference, transform the predictions back with the same function to get the actual prediction.