How to set class weights for imbalanced classes in Keras?

Dataset flower_photos.tgz

Count the number of images in different categories

import numpy as np
import tensorflow as tf

directory = 'flower_photos'
datagen = tf.keras.preprocessing.image.ImageDataGenerator()
data = datagen.flow_from_directory(directory)
unique = np.unique(data.classes, return_counts=True)
labels_dict = dict(zip(unique[0], unique[1]))
print(labels_dict)
# Found 3670 images belonging to 5 classes.
# {0: 633, 1: 898, 2: 641, 3: 699, 4: 799}

Calculate the weights of different categories

import math


def get_class_weight(labels_dict):
    """Calculate the weights of different categories

    >>> get_class_weight({0: 633, 1: 898, 2: 641, 3: 699, 4: 799})
    {0: 1.0, 1: 1.0, 2: 1.0, 3: 1.0, 4: 1.0}
    >>> get_class_weight({0: 5, 1: 78, 2: 2814, 3: 7914})
    {0: 7.366950709511269, 1: 4.619679795255778, 2: 1.034026384271035, 3: 1.0}
    """
    total = sum(labels_dict.values())
    max_num = max(labels_dict.values())
    mu = 1.0 / (total / max_num)
    class_weight = dict()
    for key, value in labels_dict.items():
        score = math.log(mu * total / float(value))
        class_weight[key] = score if score > 1.0 else 1.0
    return class_weight


labels_dict = {0: 633, 1: 898, 2: 641, 3: 699, 4: 799}
print(get_class_weight(labels_dict))
# {0: 1.0, 1: 1.0, 2: 1.0, 3: 1.0, 4: 1.0}
labels_dict = {0: 5, 1: 78, 2: 2814, 3: 7914}
print(get_class_weight(labels_dict))
# {0: 1.0, 1: 3.749820767859636, 2: 1.0, 3: 3.749820767859636, 4: 1.0, 5: 2.5931008483842453, 6: 1.0, 7: 2.5931008483842453}

If tf dataset is used you cannot use the class_weights parameter. Insted return the weight from a parse_function in your pipeline

weight_arr = [1.5, 0.5] #define your custom weights
    
#create a lookup table
key_tensor = tf.constant(list(range(0, len(weight_arr))), dtype=tf.int64)
val_tensor = tf.constant(weight_arr)
init = tf.lookup.KeyValueTensorInitializer(key_tensor, val_tensor)
weight_table = tf.lookup.StaticHashTable(init,default_value=-1)

def parse_function(element):
    features = element{'image'}
    label_integer = element{'label'}

    weight = weight_table.lookup(label_integer) #find the weight based on label

    return features, label_integer, weight

ds = ds.map(parse_function)
model.fit(ds)...

First you create a lookup table, which maps the given label integer to class weight. Then you fetch the weight based on your label in the pipeline.

I use this kind of rule for class_weight :

import numpy as np
import math

# labels_dict : {ind_label: count_label}
# mu : parameter to tune 

def create_class_weight(labels_dict,mu=0.15):
    total = np.sum(list(labels_dict.values()))
    keys = labels_dict.keys()
    class_weight = dict()
    
    for key in keys:
        score = math.log(mu*total/float(labels_dict[key]))
        class_weight[key] = score if score > 1.0 else 1.0
    
    return class_weight

# random labels_dict
labels_dict = {0: 2813, 1: 78, 2: 2814, 3: 78, 4: 7914, 5: 248, 6: 7914, 7: 248}

create_class_weight(labels_dict)

math.log smooths the weights for very imbalanced classes ! This returns :

{0: 1.0,
 1: 3.749820767859636,
 2: 1.0,
 3: 3.749820767859636,
 4: 1.0,
 5: 2.5931008483842453,
 6: 1.0,
 7: 2.5931008483842453}

Here's a one-liner using scikit-learn:

from sklearn.utils import class_weight
class_weights = dict(zip(np.unique(y_train), class_weight.compute_class_weight('balanced', np.unique(y_train), 
                y_train))) 

class_weight is fine but as @Aalok said this won't work if you are one-hot encoding multilabeled classes. In this case, use sample_weight:

sample_weight: optional array of the same length as x, containing weights to apply to the model's loss for each sample. In the case of temporal data, you can pass a 2D array with shape (samples, sequence_length), to apply a different weight to every timestep of every sample. In this case you should make sure to specify sample_weight_mode="temporal" in compile().

sample_weights is used to provide a weight for each training sample. That means that you should pass a 1D array with the same number of elements as your training samples (indicating the weight for each of those samples).

class_weights is used to provide a weight or bias for each output class. This means you should pass a weight for each class that you are trying to classify.

sample_weight must be given a numpy array, since its shape will be evaluated.

See also this answer.

I found the following example of coding up class weights in the loss function using the minist dataset. See link here.

def w_categorical_crossentropy(y_true, y_pred, weights):
    nb_cl = len(weights)
    final_mask = K.zeros_like(y_pred[:, 0])
    y_pred_max = K.max(y_pred, axis=1)
    y_pred_max = K.reshape(y_pred_max, (K.shape(y_pred)[0], 1))
    y_pred_max_mat = K.equal(y_pred, y_pred_max)
    for c_p, c_t in product(range(nb_cl), range(nb_cl)):
        final_mask += (weights[c_t, c_p] * y_pred_max_mat[:, c_p] * y_true[:, c_t])
    return K.categorical_crossentropy(y_pred, y_true) * final_mask

If you are talking about the regular case, where your network produces only one output, then your assumption is correct. In order to force your algorithm to treat every instance of class 1 as 50 instances of class 0 you have to:

1. Define a dictionary with your labels and their associated weights

   class_weight = {0: 1.,
                   1: 50.,
                   2: 2.}
   

2. Feed the dictionary as a parameter:

   model.fit(X_train, Y_train, nb_epoch=5, batch_size=32, class_weight=class_weight)
   

EDIT: "treat every instance of class 1 as 50 instances of class 0" means that in your loss function you assign higher value to these instances. Hence, the loss becomes a weighted average, where the weight of each sample is specified by class_weight and its corresponding class.

From Keras docs:

class_weight: Optional dictionary mapping class indices (integers) to a weight (float) value, used for weighting the loss function (during training only).

from collections import Counter
itemCt = Counter(trainGen.classes)
maxCt = float(max(itemCt.values()))
cw = {clsID : maxCt/numImg for clsID, numImg in itemCt.items()}

This works with a generator or standard. Your largest class will have a weight of 1 while the others will have values greater than 1 depending on how infrequent they are relative to the largest class.

Class weights accepts a dictionary type input.

Adding to the solution at https://github.com/keras-team/keras/issues/2115. If you need more than class weighting where you want different costs for false positives and false negatives. With the new keras version now you can just override the respective loss function as given below. Note that weights is a square matrix.

from tensorflow.python import keras
from itertools import product
import numpy as np
from tensorflow.python.keras.utils import losses_utils

class WeightedCategoricalCrossentropy(keras.losses.CategoricalCrossentropy):

    def __init__(
        self,
        weights,
        from_logits=False,
        label_smoothing=0,
        reduction=losses_utils.ReductionV2.SUM_OVER_BATCH_SIZE,
        name='categorical_crossentropy',
    ):
        super().__init__(
            from_logits, label_smoothing, reduction, name=f"weighted_{name}"
        )
        self.weights = weights

    def call(self, y_true, y_pred):
        weights = self.weights
        nb_cl = len(weights)
        final_mask = keras.backend.zeros_like(y_pred[:, 0])
        y_pred_max = keras.backend.max(y_pred, axis=1)
        y_pred_max = keras.backend.reshape(
            y_pred_max, (keras.backend.shape(y_pred)[0], 1))
        y_pred_max_mat = keras.backend.cast(
            keras.backend.equal(y_pred, y_pred_max), keras.backend.floatx())
        for c_p, c_t in product(range(nb_cl), range(nb_cl)):
            final_mask += (
                weights[c_t, c_p] * y_pred_max_mat[:, c_p] * y_true[:, c_t])
        return super().call(y_true, y_pred) * final_mask

You could simply implement the class_weight from sklearn:

1. Let's import the module first

   from sklearn.utils import class_weight
   

2. In order to calculate the class weight do the following

   class_weights = class_weight.compute_class_weight('balanced',
                                                    np.unique(y_train),
                                                    y_train)
   

3. Thirdly and lastly add it to the model fitting

   model.fit(X_train, y_train, class_weight=class_weights)
   

Attention: I edited this post and changed the variable name from class_weight to class_weights in order to not to overwrite the imported module. Adjust accordingly when copying code from the comments.