ACTIVATION

FUNCTION
 • An activation function are mathematical functions that
determine the output of neural network
• It is attached to each neuron in neural network and
determine whether the neuron should be
activated(“Fired”) or not

ACTIVATION • It introduce non-linear properties into network.

• An activation function is just a simple function that changes
FUNCTION its inputs into outputs with a defined range.
• Activation Function also called as transfer functions, to
transform the summed weighted input from the node into
an output value to be fed to the next hidden layer or as
output
Activation for Hidden Layers
• Activation function is one of the building block of neural network.
• When Y= ∑wixi +b then y can be anything from -∞ to + ∞. The neuron doesn’t know how to
bound the value and thus not able to decide the firing pattern.
• Activation function basically decide whether a neuron should be activated or not by
bounding value of y.
Activation Function
• All hidden layers usually use the same activation function. However, the output layer will
typically use a different activation function from the hidden layers. The choice depends on the
goal or type of prediction made by the model.
• Need of Activation Function
To add non-linearity to the neural network.
Concept of Gradient Descent
Binary ,Multiclass, Multilabel Classification
In binary classification problem statements, any of the samples from the dataset takes only one label out of
two classes.

In multi-class classification, each input will have only one output class,

In multi-label classification, each input can have multi-output classes.

Binary Class

Classify the review as either

positive or negative i.e, only
two possible target outcomes.

In neural networks, we usually use the Sigmoid Activation Function for binary classification tasks
Multi-Class Classification
For example, If we are making an animal classifier that classifies between Dog, Rabbit, Cat, and Tiger,
it makes sense only for one of these classes to be selected each time.

To ensure only one class is selected each time, we apply

the Softmax Activation Function at the last layer and we
use log loss to train the model.

Movie Review dataset

Multi-label Classification
In certain problems, each input can have multiple, or even none, of the designated output classes. In these
cases, we go for the multi-label classification problem approach.

For example, If we are building a model which predicts all the clothing articles a person is wearing, we can
use a multi-label classification model since there can be more than one possible option at once.

we use multiple sigmoids on the last layer

and thus learn a separate distribution for
each class.
Types of activation function
• Step Function
• Signum Function
• Linear Function
• Sigmoid(Logistic Function)
• ReLU (Rectified Linear Unit) Function
• Hyperbolic tan (tan h)
Binary Step Function
Input fed is compared to certain threshold, if the input is greater than it, then neuron is
activated else it is deactivated i.e. output is not passed to next layer.

Used in: Hidden layer and output layer for binary classification problems
Limitation
• It can’t provide multi value output i.e. not suitable for multiclass classification problem.
Signum Function

Used in: Hidden layer and output layer for binary classification problems
Linear Activation Function
• Also called no activation or identity function
• Activation is proportional to the input.

Range
-∞ to +∞

Used in Hidden layer or output layer for regression problem

Limitation
• It is not possible to use backpropagation as the derivative of function is a constant and has
no relation to the input x.
• All the layers of the neural network will collapse into one if a linear function is used ,no matter
the number of layers in neural network, the last layer will be linear function of first layer.
Sigmoid / Logistic Activation Function
• This function takes any real value as input and outputs values in the range of 0 to 1.
• The larger the input (more positive), the closer the output value will be to 1.0, whereas the
smaller the input (more negative), the closer the output will be to 0.0, as shown below.
Advantages:
• It is commonly used for models where we have to predict the probability as an
output. Since probability of anything exists only between the range of 0 and 1, sigmoid
is the right choice because of its range.
• Used in hidden layer, output layer for classification, tells likelihood of classification
rather than hard classification
• The function is differentiable and provides a smooth gradient, i.e., preventing jumps in
output values. This is represented by an S-shape of the sigmoid activation function.
 Limitation
• The derivative of the function is f'(x) = sigmoid(x)*(1-sigmoid(x)).

• the gradient values are only significant for range -3 to 3, and the graph gets much flatter in other regions.
It implies that for values greater than 3 or less than -3, the function will have very small gradients. As the
gradient value approaches zero, the network ceases to learn and suffers from the Vanishing
gradient problem.
• The output of the logistic function is not symmetric around zero. So the output of all the neurons will be of the
same sign. This makes the training of the neural network more difficult and unstable.
Tanh Function (Hyperbolic Tangent)
• Tanh function is very similar to the sigmoid/logistic activation function, and even has the
same S-shape with the difference in output range of -1 to 1. In Tanh, the larger the
input (more positive), the closer the output value will be to 1.0, whereas the smaller the
input (more negative), the closer the output will be to -1.0.
Advantages of using this activation function
are:

• The output of the tanh activation function is Zero centered; hence we can easily map
the output values as strongly negative, neutral, or strongly positive.
• Used in hidden layer, output layer for classification, tells likelihood of classification rather
than hard classification
• Usually used in hidden layers of a neural network.
Limitation
It also faces the problem of vanishing gradients similar to the sigmoid activation
function. Plus the gradient of the tanh function is much steeper as compared to the
sigmoid function.

• Note: Although both sigmoid and tanh face vanishing gradient issue, tanh is
zero centered, and the gradients are not restricted to move in a certain direction.
Therefore, in practice, tanh nonlinearity is always preferred to sigmoid nonlinearity.
RelU Function
• ReLU stands for Rectified Linear Unit.

• Although it gives an impression of a linear function, ReLU has a derivative function and allows for
backpropagation while simultaneously making it computationally efficient.
• Used in hidden layer of CNN or vision applications and in output layer where dependent variable
is always positive

• The main catch here is that the ReLU function does not activate all the neurons at the same time.

• The neurons will only be deactivated if the output of the linear transformation is less than 0.
def ReLU(x):
if x>0:
return x
else:
return 0

Range def relu(x):

return max(0.0, x)
[0 , ∞)
Advantages of using ReLU as an activation function are as follows:

• Since only a certain number of neurons are activated, the ReLU function is far more computationally
efficient when compared to the sigmoid and tanh functions.
• ReLU accelerates the convergence of gradient descent towards the global minimum of the loss
function due to its linear, non-saturating property.

The limitations faced by this function are:

The Dying ReLU problem.

The Dying ReLU problem
• The negative side of the graph makes the gradient value zero. Due to this reason, during the
backpropagation process, the weights and biases for some neurons are not updated. This can
create dead neurons which never get activated.
• All the negative input values become zero immediately, which decreases the model’s ability to
fit or train from the data properly.
Leaky ReLU Function
• Leaky ReLU is an improved version of ReLU function to solve the Dying ReLU problem as it has a
small positive slope in the negative area

Range
-∞ to +∞

• The advantages of Leaky ReLU are same as that of ReLU, in addition to the fact that it does enable
backpropagation, even for negative input values. Gradient is a non zero value,no dead neurons.
• Mostly used in hidden layers of CNN
Limitations
• Predictions may not be consistent for negative input values
• The gradient for negative value is small value that makes the learning of model parameter time
consuming
 Parametric ReLU Function
• Solve the problem of gradient’s becoming zero for the left half of the axis.
• This function provides the slope of the negative part of the function as an argument a. By performing
backpropagation, the most appropriate value of a is learnt.

Where "a" is the slope parameter for negative values.

• The parameterized ReLU function is used when the leaky ReLU function still fails at solving the problem
of dead neurons, and the relevant information is not successfully passed to the next layer.
Limitation
• It may perform differently for different problems depending upon the value of slope parameter a.
Softmax Function
Building Block of softmax function is sigmoid/logistic activation function that works on circulating
probability values.

The output of the sigmoid function was in the range of 0 to 1,

which can be thought of as probability.
But—
This function faces certain problems.

Suppose we have five output values of 0.8, 0.9, 0.7, 0.8, and
0.6, respectively. How can we move forward with it?

The answer is: We can’t.

The above values don’t make sense as the sum of all the
classes/output probabilities should be equal to 1.
The Softmax function is described as a combination of multiple sigmoid.

• It calculates the relative probabilities. Similar to the sigmoid/logistic activation function,

the SoftMax function returns the probability of each class.

• It is most commonly used as an activation function for the last layer of the neural network
in the case of multi-class classification.

• Mathematically it can be represented as:

• z is the vector of raw outputs from the

neural network

• The value of e ≈ 2.718

Let’s go over a simple example together.
• Assume that you have three classes, meaning that there would be three neurons in
the output layer. Now, suppose that your output from the neurons is [1.8, 0.9, 0.68].
• Applying the softmax function over these values to give a probabilistic view will
result in the following outcome: [0.58, 0.23, 0.19].
• The function returns 1 for the largest probability index while it returns 0 for the other
two array indexes. Here, giving full weight to index 0 and no weight to index 1 and
index 2. So the output would be the class corresponding to the 1st neuron(index 0)
out of three.
• You can see now how softmax activation function make things easy for multi-class
classification problems.
Activation Function for hidden layers
•Rectified Linear Activation (ReLU)
•Logistic (Sigmoid)
•Hyperbolic Tangent (Tanh)

Activation Function for output layers

• Linear
• Logistic (Sigmoid)
• Softmax
More about Activation Function?
1.ReLU activation function are more suitable for hidden layers.
2.Sigmoid/Logistic and Tanh functions should be avoided in hidden
layers as they make the model more susceptible to problems
during training (due to vanishing gradients).
3.Swish function is used in neural networks having a depth greater
than 40 layers.
Finally, a few rules for choosing the activation function for your output layer based on
the type of prediction problem that you are solving:

1.Regression - Linear Activation Function

2.Binary Classification—Sigmoid/Logistic Activation Function
3.Multiclass Classification—Softmax
4.Multilabel Classification—Sigmoid

The activation function used in hidden layers is typically chosen based on the type of
neural network architecture.

5.Convolutional Neural Network (CNN) AND Multilayer perceptron: ReLU activation

function.
6.Recurrent Neural Network: Tanh and/or Sigmoid activation function.
Neural Networks Activation Functions in a Nutshell
• Activation Functions are used to introduce non-linearity in the network.

• A neural network will almost always have the same activation function in all hidden layers. This
activation function should be differentiable so that the parameters of the network are learned in
backpropagation.

• ReLU is the most commonly used activation function for hidden layers.

• While selecting an activation function, you must consider the problems it might face: vanishing
and exploding gradients.

• Regarding the output layer, we must always consider the expected value range of the predictions.
If it can be any numeric value (as in case of the regression problem) you can use the linear
activation function or ReLU.

• Use Softmax or Sigmoid function for the classification problems.

How to choose an output layer activation
function