Loss function for class imbalanced binary classifier in Tensor flow
I am trying to apply deep learning for a binary classification problem with high class imbalance between target classes (500k, 31K). I want to write a custom loss function which should be like: minimize(100-((predicted_smallerclass)/(total_smallerclass))*100) Appreciate any pointers on how I can build this logic.
You can add class weights to the loss function, by multiplying logits. Regular cross entropy loss is this: loss(x, class) = -log(exp(x[class]) / (\sum_j exp(x[j]))) = -x[class] + log(\sum_j exp(x[j])) in weighted case: loss(x, class) = weights[class] * (-x[class] + log(\sum_j exp(x[j]))) So by multiplying logits, you are re-scaling predictions of each class by its class weight. For example: ratio = 31.0 / (500.0 + 31.0) class_weight = tf.constant([ratio, 1.0 - ratio]) logits = ... # shape [batch_size, 2] weighted_logits = tf.mul(logits, class_weight) # shape [batch_size, 2] xent = tf.nn.softmax_cross_entropy_with_logits( weighted_logits, labels, name="xent_raw") For a complete solution, you can use this op in skflow: https://github.com/tensorflow/tensorflow/blob/master/tensorflow/contrib/learn/python/learn/ops/losses_ops.py
The code you proposed seems wrong to me. The loss should be multiplied by the weight, I agree. But if you multiply the logit by the class weights, you end with: weights[class] * -x[class] + log( \sum_j exp(x[j] * weights[class]) ) The second term is not equal to: weights[class] * log(\sum_j exp(x[j])) To show this, we can be rewrite the latter as: log( (\sum_j exp(x[j]) ^ weights[class] ) So here is the code I'm proposing: ratio = 31.0 / (500.0 + 31.0) class_weight = tf.constant([[ratio, 1.0 - ratio]]) logits = ... # shape [batch_size, 2] weight_per_label = tf.transpose( tf.matmul(labels , tf.transpose(class_weight)) ) #shape [1, batch_size] # this is the weight for each datapoint, depending on its label xent = tf.mul(weight_per_label , tf.nn.softmax_cross_entropy_with_logits(logits, labels, name="xent_raw") #shape [1, batch_size] loss = tf.reduce_mean(xent) #shape 1
Use tf.nn.weighted_cross_entropy_with_logits() and set pos_weight to 1 / (expected ratio of positives).
Did ops tf.nn.weighted_cross_entropy_with_logits() for two classes: classes_weights = tf.constant([0.1, 1.0]) cross_entropy = tf.nn.weighted_cross_entropy_with_logits(logits=logits, targets=labels, pos_weight=classes_weights)
Different results in Weka GUI and Weka via Java code
imbalanced data classification with boosting algorithms
How to create ARFF file for 2D data points?
How to use weighted vote for classification using weka
Convert Web page to ARFF File for Weka classification
Liblinear bias greater than 2 improving accuracy?
Weka: Does training helps if test run is followed by training run?
Difference between logistic regression with binary output and classification
Weka - How to find input format for classifiers
How to incorporate Weka Naive Bayes model into Java Code
RapidMiner: Classifying new examples without re-running the existing trained model
How to check whether data is being overfiited for that model in weka
Feature Extraction for Face Dectection
rapid-miner formating datsets with many parameter
text classification methods? SVM and decision tree
Multilabel classification with SVM using rapidminer