Perplexing Dilemma: AI ML DL DS

Original article was published by TechBuds on Artificial Intelligence on Medium

Perplexing Dilemma: AI ML DL DS

Feel the difference: AI vs ML vs DL vs DS

These are the terms that have demented a lot of people. So if you are one of them, then this blog will clear all your confusion. We are going to tell you what exactly these terms are and how we use Data Science considering all these particular technologies and work.


AI is the biggest umbrella comprising ML and Deep learning.

AI enables the machine to act like humans by replicating behavior and nature.

With AI it is possible for machines to learn from past experience and develop with time.

AI can be trained to accomplish specific tasks by processing a large amount of data and recognizing a pattern in them.

This concept was built in the 1940s But has built its popularity recently.


Prior to this, we had only limited data which was not sufficient for the machine learning and deep learning models to predict accurate results but now there is a tremendous increase in the amount of data that results in better predictions.

It enables the computer or machine to think without any human intervention and thus is capable of taking its own decisions.

So, Our final goal is to derive an AI application.

One of the most common AI applications is our apple series just playing compute Tesla self-driving car.

What is an AI app?

It is nothing but an app that uses ML and keeps learning to perform specific tasks.


ML is a subset of AI. ML provides us statistical tools to explore, analyze, and understand that data.

We have 3 techniques for the same.

1. Supervised Learning:

In supervised Learning, a labeled dataset is used for making predictions. For instance, Consider height and weight as 2 features and we have to classify whether this person belongs to the obese category or the fit category. So with the help of the present, we can train our model.

2. Unsupervised Learning:

In Unsupervised learning, there is no labeled data i.e, we don’t know the output.

Generally, clustering techniques are used to build the model. We have different clustering techniques like K means, DB Scan clustering for the same.

Clustering exactly means a grouping of data based on the similarity criteria. There are many mathematical concepts like euclidean distance used along with some other weight reduction and techniques.

3. Reinforcement learning (semi-supervised):

In Reinforcement Learning, some parts of the past data are known and it learns more as a new environment and the new dataset is provided.


It can be described as the extended version of machine learning.

Earlier the biggest question in the world was?

Can machines think and learn like humans? If anybody had asked us that question in the late ’80s or early ’90s, we would have said no but today it is possible with the help of Deep Learning.

The whole idea behind was to create a structure that can mimic the human brain.

Here we create an architecture called the Multi neural network architecture.

These models actually learn how a human brain learns.

To tackle different complex problems, techniques like ANN, CNN and RNN are used

1. ANN (Artificial Neural Network) :

Most of the data that is presented in the form of numbers is solved by ANN.

2. CNN(Convolution Neural Network):

If our input is in the form of images we will use CNN.

3. RNN(Recurrent Neural network)

Suppose our input is in the form of a tiny series kind of data at that time we will be using RNN.


So now the final question arises…. Where does Data Science fit?

Data Science is the technique that tries to apply all the above models. It makes use of many mathematical tools like statistics, probability, linear algebra, Differential Calculus, etc.

A data scientist may have to work on Machine learning techniques as well as Deep Learning techniques based on the case study along with the application of these tools.

Isn’t it interesting to build our own self-driving or to build a recommendation system on our own?