How Machine Learning and Deep Learning is related to Data Science

Source: Deep Learning on Medium

How Machine Learning and Deep Learning is related to Data Science

Figure Showing how Data Science is related to Artificial Intelligence
Source: www.ni.com

As can be seen from the above block diagram, Data science is a different domain of data handling that relates itself to the class of Artificial Intelligence by the use of various machine learning and deep learning algorithms for the exploratory analysis and prediction.

Let’s understand what exactly these terms are.

ARTIFICIAL INTELLIGENCE(AI):

System with an intelligence, i.e. with a decision making capability can be said as an artificially intelligent system.

Source: http://sitn.hms.harvard.edu/flash/2017/history-artificial-intelligence/

As can be seen from the above figure, the history of AI lies from long before. May be Alan Turing was the first to state that,” Machine can think”.The classical AI is mainly based on helping the computer for decision making with the help of some logic. Typical AI systems include:

  1. Expert Systems(works on a certain rule base)
  2. Graph Search Mechanisms(Depth First Search and Breadth-First Search)

With the hype in digital communications and internet compatibility, in the late ’90s, a huge amount of data getting generated. So presently in 2020, modern AI is almost completely data-driven. Machine Learning and Deep Learning are the part of AI that solves the given problem by using data as the source.

MACHINE LEARNING (ML):

Computers making decisions without being explicitly programmed.

Source: towards data science.com

The term Machine Learning is basically a short-hand for Traditional Machine Learning. With the help of ML, the system is made to solve the decision making task without being explicitly programmed.

In ML pipeline, the programmer manually selects the features of a data and feeds them to the machine for training by using some algorithms like Support Vector Machine(SVM), Ensemble Methods, Decision Trees, etc.

Broadly, ML algorithms can be classified as

  1. Supervised Learning
  2. Unsupervised Learning
  3. Reinforcement Learning

In Supervised learning, the user has knowledge of the ground truth. Whereas in unsupervised learning methods, the machine itself categorizes the given data into different sets and makes a decision. The concept of Reinforcement Learning is however based on a reward- penanlty mechanism. It is a special type of unsupervised learning.

DEEP LEARNING(DL):

Computers making decisions where decision criteria need not to be defined or guided.

Figure showing an application of deep learning

It is basically a subset of machine learning algorithms, that mainly uses different neural network models, to train a system with the given data and to answer the following questions.

  1. Whether any object or pattern is detected in the given data?
  2. What is the type of detected object? Is it desired or undesired?
  3. Where is it detected?

To answer these questions, a neural network needs to be trained with sufficient input data for a sufficient amount of time. After training, the model gets enabled for classification, detection, segmentation, prediction purpose, etc.

Typical Deep learning models include

  1. convolutional Neural network(CNN)
  2. recurrent Neural network(RNN)
  3. Models for Semantic Segmentation etc.

DATA SCIENCE(DS):

Data Science is quite broader than just learning the data. It is not always necessary to make the machine learn with the data to solve a certain task.

Photo by Franki Chamaki on Unsplash

But it intersects with AI by using data-driven ML and DL models for analyzing and describing the given data. In other words, it can be interpreted that, data science use ML and DL as a tool for analyzing the data.

It includes collecting, storing, processing, describing and modeling the data. Knowledge in statistics and mathematics may be helpful for this purpose.

A detailed discussion on data science can be found at: