ML1. Introduction to Machine Learning

ML1. Introduction to Machine Learning

Introduction

In 1959, Arthur Samuel defined machine learning as follows:

“Field of study that gives computers the ability to learn without being explicitly programmed”

"Explicitly programmed" in this case means that we don't have to code everything for our machine to do a task. We will show our machine some examples, and then it will figure things out on its own.

Examples

Now we will see some examples of machine learning that we use in our daily life maybe without even knowing it.

Virtual Assistants

Smartphone assistants like Apple Siri use machine learning to recognize speech, answer questions and do other smart things. Assistants like Siri and Google Assistant are powered by automatic speech recognition and Natural Language Processing (NLP).

AI Assistant Siri

Self Driving Car

One of the most exciting and cutting-edge uses of machine learning algorithms is in autonomous vehicles. Self-driving cars can significantly reduce traffic, and most importantly they can reduce road accidents

Self Driving Car from Tesla

Spam or Fraud Detection

Machine learning is used in every spam filter, such as in Gmail.

Gmail Spam Detection

ML systems are also used by credit card companies and banks to automatically detect fraudulent behavior.

How Machines Learn?

There are two types of ways our machine can learn:

Supervised Learning

  • Used most in real-world applications

  • Rapid advancement

In this type of learning, we give the machine some input and some labeled output. This input data will help the machine train itself so that it can predict the outcome as accurately as possible as soon as it gets input. In the Coursera video, we saw an example of regression where the task is to predict a number. There is also a second major type of supervised learning which is called classification.

Unsupervised Learning

In this type of learning, we give the machine some data that is not tagged or labeled. The idea is that the computer will be compelled to construct a succinct representation through mimicry, a key way of learning in humans, and then use that for creative output.

Why Machine Learning?

  • It allows for building practical systems for real-world applications that couldn't be solved otherwise.

  • Learning is widely regarded as a key approach to building general-purpose artificial intelligence systems.

  • The science and engineering of machine learning offer insights into human intelligence.