Neural Networks: Chapter 3 - Layers ...

Neural Networks Jun 11, 2021

In our previous posts we took a look at What are Neural Networks? and it's capabilities. Now that we have a better understanding of what they are, let's start peeling the Onion.

In today's article we will answer some of the very fundamental questions of Neural Networks. Like What encompasses it? How are they built? How do they work? What do we need? and a lot more.

As we progress further we will keep diving in deeper into the realm of Neural Networks. So buckle up and Enjoy the Ride.

What makes a Neural Network?

As mentioned in our previous articles a NODE is the most fundamental unit of a Neural Network.
A collection of NODES with the same FUNCTIONALITY and Activation Function scaled across vertically define a LAYER.

Multiple Layers stacked together form what we call Neural Network. Let's take a look at What do they actually look like.

As it is clearly visible from the above video, that a very primitive Neural Network comprises of a complex web of NODES and LAYERS. Where each node and layer have their own functionality in terms of FEATURE EXTRACTION.

What are Layers?

As we mentioned above, Layers are an aggregation of multiple nodes. But then why do we stress so much on the concept of layers?

We need not run through in details with every concept, but let's try and have an intuitive understanding of their impact.

So imagine yourself as a Teacher to say 50 students. Now if some of your students fail in an exam, it is your duty to provide feedback to the students and re-train them to perform better in the next exam.
But what if you catered to students belonging to different age groups. In such a case it wouldn't be feasible to re-train and provide feedback to each individual. In order to solve such a problem, instinctively one would club all the students of a single age group and have extra classes for them.

That way a single teacher would be able to cater to multiple students and provide significant but normalised feedback and training to all. This concept is the reason why we have Nodes aggregated into Layers.

  1. For easier training
  2. For smoother feedback and updates
  3. For faster processing
  4. For standardisation

What are the type of Layers?

As a part of our reading process, we have got an intuitive understanding of what Neural Architectures are and How AI can be implemented in the society.

Now for an obvious statement - Humans have 5 basic senses to help interact with the environment

In the same manner if we are building an AI, it needs to have the capability to cater to multiple DATA types, encompassing all means of Interactions and Stimuli. In order to tackle different types of data, researchers have developed multiple variety of layers specialised in tackling particular tasks and datasets.


Now that we have a rough idea of What Layers are and What they can leveraged for, along with their variations. I would recommend if you all would go ahead and explore it for yourself.

Click on the LINK to explore Neural Networks in 3D

In our future posts we will be taking a deeper dive into each of the Layers mentioned above and How we can implement them into our AI solutions.



Vaibhav Satpathy

AI Enthusiast and Explorer

Great! You've successfully subscribed.
Great! Next, complete checkout for full access.
Welcome back! You've successfully signed in.
Success! Your account is fully activated, you now have access to all content.