Why Deep Learning uses the term "Hidden Layers"?

.

The term "hidden layer" in neural networks refers to the intermediate layers between the input and output layers. 

While the term "hidden layer" might seem misleading at first, there are some reasons why the term 'hidden layers' persists in the context of neural networks:

(1) Historical Context:

The term originated decades ago when neural networks were simpler and had only one hidden layer due to computational limitations.

Researchers referred to this layer as "hidden" because its internal computations were not explicitly visible.

(2) Mathematical Interpretation:

Each neuron in a hidden layer computes a weighted sum of inputs and applies an activation function.

These intermediate computations are not directly exposed to the user or external observer.

From a mathematical perspective, they remain "hidden".

(3) Functionality and Abstraction:

Hidden layers perform essential computations within the neural network.

They transform input data into higher-level representations.

Despite being technically accessible, their purpose is abstracted away for simplicity.

(4) Analogy to Brain Neurons:

Neural networks draw inspiration from the human brain.

Just as we cannot directly observe individual brain neurons’ inner workings, the computations in hidden layers remain "hidden".

(5) Deep Learning and Stacking Layers:

Modern deep learning architectures involve stacking multiple hidden layers.

Each layer learns increasingly abstract features.

The term "deep" refers to the depth (number of layers) in these networks.

In summary, while we understand the properties of hidden layers, the term endures as a historical artifact and a nod to the network’s origins. 

It reminds us that powerful transformations occur within these layers, even if they are no longer truly "hidden". 

.