What is Machine Learning? Need of ML.

What is Machine Learning?

Rolof Computer Academy

• Machine Learning is a subset of Artificial Intelligence.

Arthur Samuel defined Machine Learning in 1959 as

The field of study that gives computers the ability to learn without being explicitly programmed.

• So Machine Learning is not a new concept it's from decades, but we are using it more now.
• Why now? not earlier, because due to digitization we are having a lot of data and new data is generated every day and machine learning needs a lot of data.

How does Machine Learning work? And the need for it.

• Machine Learning is a set of algorithms, which can learn with proper data without the use of traditional programming.

Famous example: Spam Classification

• Here we need to classify whether a mail is a spam or not.

Machine Learning

• The user provides the algorithm with a large number of emails (inputs), together with information on whether emails are spam or not.
• The algorithm will then produce a prediction as to whether the new email is spam or not.

Traditional Programming

• A Human expert should design a list of hardcoded rules using “if” and “else” to process data and could make up a blacklist of words that would result in an email being marked as spam or not

• As we can see from above the traditionally programming sucks !!
• And when a new dataset is formed and for it, new rules should be decided, hardcoded and it's a repetitive time-consuming process.
• Here Machine Learning comes to the rescue.

Machine Learning Algorithm takes input data and “generates rules” so that when an unknown input is passed in it we get predicted output.

Passed in what? 😐

In mapping function

Here,

• Y is Output (also called Output Variable / Dependent Variable)
• X is Input (also called Input Variables / Input Vector / Independent Variable)
• f() is mapping function (also called Algorithm)

Types of Machine Learning

Based on X (Input Variables) and Y (Output Variables) machine learning algorithms are classified in three categories :

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

1. Supervised Learning

Image by tutorialandexample

• In supervised learning mapping function f() maps an Input variable (X) to an Output variable (Y) based on input — output pairs.
• Each example in training dataset is pair consisting of an input vector object and an output value.
• Supervised learning algorithms are trained using labeled data. The model predicts the output.
• Further supervised learning can be categorized in Classification and Regression problems.

2. Unsupervised Learning

• In Unsupervised learning we have only input data and no corresponding output variable.
• The main aim of unsupervised learning algorithm is to find the underlying structure or distribution in the data.
• By using unsupervised learning we can find patterns based on features of input data.
• Further unsupervised learning can be categorized in Clustering and Association problems.

3. Reinforcement Learning

Image by mathworks

• Reinforcement learning is a trial-and-trial approach to learning. For every action the machine takes, it receives a Reward or Penalty point.
• The machine receives a reward point if the option is accurate, or a penalty point if the response is incorrect.
• The interaction between the environment and the learning agent is important to the reinforcement learning algorithm.
• The agent develops its surroundings knowledge in order to pick or conduct the next action, with the goal of earning additional rewards.

Conclusions

In this article covered shortly about machine learning, need of it and compared it with traditional programming and at last types of machine learning.

This is my first article for any feedback mail me..