Difference between ReLU, ELU and Leaky ReLU. Their pros and cons majorly

Question

I am unable to understand when to use ReLU, ELU and Leaky ReLU. How do they compare to other activation functions(like the sigmoid and the tanh) and their pros and cons.

Answer 1

Look at this ML glossary:

ELU

ELU is very similiar to RELU except negative inputs. They are both in identity function form for non-negative inputs. On the other hand, ELU becomes smooth slowly until its output equal to $-\alpha$ whereas RELU sharply smoothes.

Pros

• ELU becomes smooth slowly until its output equal to $-\alpha$ whereas RELU sharply smoothes.
• ELU is a strong alternative to ReLU.
• Unlike to ReLU, ELU can produce negative outputs.

Cons

• For $x > 0$, it can blow up the activation with the output range of [0, inf].

ReLU

Pros

• It avoids and rectifies vanishing gradient problem.
• ReLu is less computationally expensive than tanh and sigmoid because it involves simpler mathematical operations.

Cons

• One of its limitations is that it should only be used within hidden layers of a neural network model.
• Some gradients can be fragile during training and can die. It can cause a weight update which will makes it never activate on any data point again. In other words, ReLu can result in dead neurons.
• In another words, For activations in the region ($x<0$) of ReLu, gradient will be 0 because of which the weights will not get adjusted during descent. That means, those neurons which go into that state will stop responding to variations in error/ input (simply because gradient is 0, nothing changes). This is called the dying ReLu problem.
• The range of ReLu is $[0,\infty)$. This means it can blow up the activation.

LeakyRelu

LeakyRelu is a variant of ReLU. Instead of being 0 when $z<0$, a leaky ReLU allows a small, non-zero, constant gradient α (Normally, $\alpha=0.01$). However, the consistency of the benefit across tasks is presently unclear. [1]

Pros

• Leaky ReLUs are one attempt to fix the “dying ReLU” problem by having a small negative slope (of 0.01, or so).

Cons

• As it possess linearity, it can’t be used for the complex Classification. It lags behind the Sigmoid and Tanh for some of the use cases.

Further reading

• Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification, Kaiming He et al. (2015)