The Situation:
I am wondering how to use TensorFlow optimally when my training data is imbalanced in label distribution between 2 labels. For instance, suppose the MNIST tutorial is simplified to only distinguish between 1's and 0's, where all images available to us are either 1's or 0's. This is straightforward to train using the provided TensorFlow tutorials when we have roughly 50% of each type of image to train and test on. But what about the case where 90% of the images available in our data are 0's and only 10% are 1's? I observe that in this case, TensorFlow routinely predicts my entire test set to be 0's, achieving an accuracy of a meaningless 90%.
One strategy I have used to some success is to pick random batches for training that do have an even distribution of 0's and 1's. This approach ensures that I can still use all of my training data and produced decent results, with less than 90% accuracy, but a much more useful classifier. Since accuracy is somewhat useless to me in this case, my metric of choice is typically area under the ROC curve (AUROC), and this produces a result respectably higher than .50.
Questions:
(1) Is the strategy I have described an accepted or optimal way of training on imbalanced data, or is there one that might work better?
(2) Since the accuracy metric is not as useful in the case of imbalanced data, is there another metric that can be maximized by altering the cost function? I can certainly calculate AUROC post-training, but can I train in such a way as to maximize AUROC?
(3) Is there some other alteration I can make to my cost function to improve my results for imbalanced data? Currently, I am using a default suggestion given in TensorFlow tutorials:
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(pred, y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)
I have heard this may be possible by up-weighting the cost of miscategorizing the smaller label class, but I am unsure of how to do this.
(1)It's ok to use your strategy. I'm working with imbalanced data as well, which I try to use down-sampling and up-sampling methods first to make the training set even distributed. Or using ensemble method to train each classifier with an even distributed subset.
(2)I haven't seen any method to maximise the AUROC. My thought is that AUROC is based on true positive and false positive rate, which doesn't tell how well it works on each instance. Thus, it may not necessarily maximise the capability to separate the classes.
(3)Regarding weighting the cost by the ratio of class instances, it similar to Loss function for class imbalanced binary classifier in Tensor flow
and the answer.
Regarding imbalanced datasets, the first two methods that come to mind are (upweighting positive samples, sampling to achieve balanced batch distributions).
Upweighting positive samples
This refers to increasing the losses of misclassified positive samples when training on datasets that have much fewer positive samples. This incentivizes the ML algorithm to learn parameters that are better for positive samples. For binary classification, there is a simple API in tensorflow that achieves this. See (weighted_cross_entropy) referenced below
https://www.tensorflow.org/api_docs/python/tf/nn/weighted_cross_entropy_with_logits
Batch Sampling
This involves sampling the dataset so that each batch of training data has an even distribution positive samples to negative samples. This can be done using the rejections sampling API provided from tensorflow.
https://www.tensorflow.org/api_docs/python/tf/contrib/training/rejection_sample
I'm one who struggling with imbalanced data. What my strategy to counter imbalanced data are as below.
1) Use cost function calculating 0 and 1 labels at the same time like below.
cost = tf.reduce_mean(-tf.reduce_sum(y*tf.log(_pred) + (1-y)*tf.log(1-_pred), reduction_indices=1))
2) Use SMOTE, oversampling method making number of 0 and 1 labels similar. Refer to here, http://comments.gmane.org/gmane.comp.python.scikit-learn/5278
Both strategy worked when I tried to make credit rating model.
Logistic regression is typical method to handle imbalanced data and binary classification such as predicting default rate. AUROC is one of the best metric to counter imbalanced data.
1) Yes. This is well received strategy to counter imbalanced data. But this strategy is good in Neural Nets only if you using SGD.
Another easy way to balance the training data is using weighted examples. Just amplify the per-instance loss by a larger weight/smaller when seeing imbalanced examples. If you use online gradient descent, it can be as simple as using a larger/smaller learning rate when seeing imbalanced examples.
Not sure about 2.
Related
I was using Keras' CNN to classify MNIST dataset. I found that using different batch-sizes gave different accuracies. Why is it so?
Using Batch-size 1000 (Acc = 0.97600)
Using Batch-size 10 (Acc = 0.97599)
Although, the difference is very small, why is there even a difference?
EDIT - I have found that the difference is only because of precision issues and they are in fact equal.
That is because of the Mini-batch gradient descent effect during training process. You can find good explanation Here that I mention some notes from that link here:
Batch size is a slider on the learning process.
Small values give a learning process that converges quickly at the
cost of noise in the training process.
Large values give a learning
process that converges slowly with accurate estimates of the error
gradient.
and also one important note from that link is :
The presented results confirm that using small batch sizes achieves the best training stability and generalization performance, for a
given computational cost, across a wide range of experiments. In all
cases the best results have been obtained with batch sizes m = 32 or
smaller
Which is the result of this paper.
EDIT
I should mention two more points Here:
because of the inherent randomness in machine learning algorithms concept, generally you should not expect machine learning algorithms (like Deep learning algorithms) to have same results on different runs. You can find more details Here.
On the other hand both of your results are too close and somehow they are equal. So in your case we can say that the batch size has no effect on your network results based on the reported results.
This is not connected to Keras. The batch size, together with the learning rate, are critical hyper-parameters for training neural networks with mini-batch stochastic gradient descent (SGD), which entirely affect the learning dynamics and thus the accuracy, the learning speed, etc.
In a nutshell, SGD optimizes the weights of a neural network by iteratively updating them towards the (negative) direction of the gradient of the loss. In mini-batch SGD, the gradient is estimated at each iteration on a subset of the training data. It is a noisy estimation, which helps regularize the model and therefore the size of the batch matters a lot. Besides, the learning rate determines how much the weights are updated at each iteration. Finally, although this may not be obvious, the learning rate and the batch size are related to each other. [paper]
I want to add two points:
1) When use special treatments, it is possible to achieve similar performance for a very large batch size while speeding-up the training process tremendously. For example,
Accurate, Large Minibatch SGD:Training ImageNet in 1 Hour
2) Regarding your MNIST example, I really don't suggest you to over-read these numbers. Because the difference is so subtle that it could be caused by noise. I bet if you try models saved on a different epoch, you will see a different result.
I have data set for classification problem. I have in total 50 classes.
Class1: 10,000 examples
Class2: 10 examples
Class3: 5 examples
Class4: 35 examples
.
.
.
and so on.
I tried to train my classifier using SVM ( both linear and Gaussian kernel). My accurate is very bad on test data 65 and 72% respectively. Now I am thinking to go for a neural network. Do you have any suggestion for any machine learning model and algorithm for large imbalanced data? It would be extremely helpful to me
You should provide more information about the data set features and the class distribution, this would help others to advice you.
In any case, I don't think a neural network fits here as this data set is too small for it.
Assuming 50% or more of the samples are of class 1 then I would first start by looking for a classifier that differentiates between class 1 and non-class 1 samples (binary classification). This classifier should outperform a naive classifier (benchmark) which randomly chooses a classification with a prior corresponding to the training set class distribution.
For example, assuming there are 1,000 samples, out of which 700 are of class 1, then the benchmark classifier would classify a new sample as class 1 in a probability of 700/1,000=0.7 (like an unfair coin toss).
Once you found a classifier with good accuracy, the next phase can be classifying the non-class 1 classified samples as one of the other 49 classes, assuming these classes are more balanced then I would start with RF, NB and KNN.
There are multiple ways to handle with imbalanced datasets, you can try
Up sampling
Down Sampling
Class Weights
I would suggest either Up sampling or providing class weights to balance it
https://towardsdatascience.com/5-techniques-to-work-with-imbalanced-data-in-machine-learning-80836d45d30c
You should think about your performance metric, don't use Accuracy score as your performance metric , You can use Log loss or any other suitable metric
https://machinelearningmastery.com/failure-of-accuracy-for-imbalanced-class-distributions/
From my experience the most successful ways to deal with unbalanced classes are :
Changing distribiution of inputs: 20000 samples (the approximate number of examples which you have) is not a big number so you could change your dataset distribiution simply by using every sample from less frequent classes multiple times. Depending on a number of classes you could set the number of examples from them to e.g. 6000 or 8000 each in your training set. In this case remember to not change distribiution on test and validation set.
Increase the time of training: in case of neural networks, when changing distribiution of your input is impossible I strongly advise you trying to learn network for quite a long time (e.g. 1000 epochs). In this case you have to remember about regularisation. I usually use dropout and l2 weight regulariser with their parameters learnt by random search algorithm.
Reduce the batch size: In neural networks case reducing a batch size might lead to improving performance on less frequent classes.
Change your loss function: using MAPE insted of Crossentropy may also improve accuracy on less frequent classes.
Feel invited to test different combinations of approaches shown by e.g. random search algorithm.
Data-level methods:
Undersampling runs the risk of losing important data from removing data. Oversampling runs the risk of overfitting on training data, especially if the added copies of the minority class are replicas of existing data. Many sophisticated sampling techniques have been developed to mitigate these risks.
One such technique is two-phase learning. You first train your model on the resampled data. This resampled data can be achieved by randomly undersampling large classes until each class has only N instances. You then fine-tune your model on the original data.
Another technique is dynamic sampling: oversample the low-performing classes and undersample the high-performing classes during the training process. Introduced by Pouyanfar et al., the method aims to show the model less of what it has already learned and more of what it has not.
Algorithm-level methods
Cost-sensitive learning
Class-balanced loss
Focal loss
References:
esigning Machine Learning Systems
Survey on deep learning with class imbalance
I am interested in any tips on how to train a set with a very limited positive set and a large negative set.
I have about 40 positive examples (quite lengthy articles about a particular topic), and about 19,000 negative samples (most drawn from the sci-kit learn newsgroups dataset). I also have about 1,000,000 tweets that I could work with.. negative about the topic I am trying to train on. Is the size of the negative set versus the positive going to negatively influence training a classifier?
I would like to use cross-validation in sci-kit learn. Do I need to break this into train / test-dev / test sets? Is know there are some pre-built libraries in sci-kit. Any implementation examples that you recommend or have used previously would be helpful.
Thanks!
The answer to your first question is yes, the amount by which it will affect your results depends on the algorithm. My advive would be to keep an eye on the class-based statistics such as recall and precision (found in classification_report).
For RandomForest() you can look at this thread which discusses
the sample weight parameter. In general sample_weight is what
you're looking for in scikit-learn.
For SVM's have a look at either this example or this
example.
For NB classifiers, this should be handled implicitly by Bayes
rule, however in practice you may see some poor performances.
For you second question it's up for discussion, personally I break my data into a training and test split, perform cross validation on the training set for parameter estimation, retrain on all the training data and then test on my test set. However the amount of data you have may influence the way you split your data (more data means more options).
You could probably use Random Forest for your classification problem. There are basically 3 parameters to deal with data imbalance. Class Weight, Samplesize and Cutoff.
Class Weight-The higher the weight a class is given, the more its error rate is decreased.
Samplesize- Oversample the minority class to improve class imbalance while sampling the defects for each tree[not sure if Sci-kit supports this, used to be param in R)
Cutoff- If >x% trees vote for the minority class, classify it as minority class. By default x is 1/2 in Random forest for 2-class problem. You can set it to a lower value for the minority class.
Check out balancing predict error at https://www.stat.berkeley.edu/~breiman/RandomForests/cc_home.htm
For the 2nd question if you are using Random Forest, you do not need to keep separate train/validation/test set. Random Forest does not choose any parameters based on a validation set, so validation set is un-necessary.
Also during the training of Random Forest, the data for training each individual tree is obtained by sampling by replacement from the training data, thus each training sample is not used for roughly 1/3 of the trees. We can use the votes of these 1/3 trees to predict the out of box probability of the Random forest classification. Thus with OOB accuracy you just need a training set, and not validation or test data to predict performance on unseen data. Check Out of Bag error at https://www.stat.berkeley.edu/~breiman/RandomForests/cc_home.htm for further study.
We all know that the objective function of SVM is iteratively trained. In order to continue training, at least we can store all the variables used in the iterations if we want to continue on the same training dataset.
While, if we want to train on a slightly different dataset, what should we do to make full use of the previously trained model? Or does this kind of thought make sense? I think it is quite reasonable if we train a K-means model. But I am not sure if it still makes sense for the SVM problem.
There are some literature on this topic:
alpha-seeding, in which the training data is divided into chunks. After you train a SVM on the ith chunk, you take those and use them to train your SVM with the (i+1)th chunk.
Incremental SVM serves as an online learning in which you update a classifier with new examples rather than retrain the entire data set.
SVM heavy package with online SVM training as well.
What you are describing is what an online learning algorithm does and unfortunately the classic definition for SVM is done in a batch fashion.
However, there are several solvers for SVM that produces quasy optimal hypothesis to the underneath optimization problem in an online learning way. In particular my favourite is Pegasos-SVM which can find a good near optimal solution in linear time:
http://ttic.uchicago.edu/~nati/Publications/PegasosMPB.pdf
In general this doesn't make sense. SVM training is an optimization process with regard to every training set vector. Each training vector has an associated coefficient, which as a result is either 0 (irrelevant) or > 0 (support vector). Adding another training vector imposes another, different, optimization problem.
The only way to reuse information from previous training I can think of is to choose support vectors from the previous training and add them to the new training set. I'm not sure, but this probably will negatively affect generalization - VC dimension of an SVM is related to the number of support vectors, so adding previous support vectors to the new dataset is likely to increase the support vector count.
Apparently, there are more possibilities, as noted in lennon310's answer.
I'm with a problem when I try to classify my data using libsvm. My training and test data are highly unbalanced. When I do the grid search for the svm parameters and train my data with weights for the classes, the testing gives the accuracy of 96.8113%. But because the testing data is unbalanced, all the correct predicted values are from the negative class, which is larger than the positive class.
I tried a lot of things, from changing the weights until changing the gamma and cost values, but my normalized accuracy (which takes into account the positive classes and negative classes) is lower in each try. Training 50% of positives and 50% of negatives with the default grid.py parameters i have a very low accuracy (18.4234%).
I want to know if the problem is in my description (how to build the feature vectors), in the unbalancing (should i use balanced data in another way?) or should i change my classifier?
Better data always helps.
I think that imbalance is part of the problem. But a more significant part of the problem is how you're evaluating your classifier. Evaluating accuracy given the distribution of positives and negatives in your data is pretty much useless. So is training on 50% and 50% and testing on data that is distributed 99% vs 1%.
There are problems in real life that are like the one your studying (that have a great imbalance in positives to negatives). Let me give you two examples:
Information retrieval: given all documents in a huge collection return the subset that are relevant to search term q.
Face detection: this large image mark all locations where there are human faces.
Many approaches to these type of systems are classifier-based. To evaluate two classifiers two tools are commonly used: ROC curves, Precision Recall curves and the F-score. These tools give a more principled approach to evaluate when one classifier is working better than the another.