Introduction to Machine Learning

Malsha Nishadini
4 min readApr 19, 2019

--

These days, Machine Learning(ML) has become a very popular topic among many IT students and IT professionals. As an IT student, I’m also very interested in Machine Learning. I’m willing to share few basic things in ML that I have learned.

What is Machine Learning?

You can see there is no well accepted definition of what is Machine Learning in the internet. But the common definition of Machine Learning is: “ the scientific study of algorithms and statistical models that computer systems use to effectively perform a specific task without being explicitly programmed.” Additionally, I would like to explain what is ML using a more recent definition by Tom Mitchell who is an American computer scientist.

Definition by Tom Mitchell:

“A computer program is said to learn from experience E with respect to some Task T and some performance measure P, if its performance on T, as measured by P, improves with experience E.”

I know, this is little difficult to understand at once. Let’s take an example to clear the definition.

Suppose your email program watches which emails you do or do not mark as spam, and based on that the email program learns how to better filter spam. In this case,

T = Classifying emails as spam or not spam

E = Watching you label emails as spam or not spam

P = The number(or fraction) of emails correctly classified as spam or not spam

Machine Learning is considered as a subset of artificial Intelligence. It is used to make predictions or decisions through mathematical model of sample data(training data) that is built by mathematical algorithms in ML. The predictions are evaluated for accuracy before deploying the machine learning algorithm. If there is no acceptable accuracy ML algorithms are trained again and again with training data.

How does Machine Learning work — Image Source:Google

Machine Learning Algorithms

Mainly, there are three types of ML algorithms and they differ in their approaches.

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

Supervised Learning

In supervised learning, we give the algorithm a training data set and consists of a set of training examples. In every example, we are told that what is the correct output for the relevant input. Then the algorithm is predicted on that example. Finally, the algorithm will be able to correctly determine the output for inputs that were not a part of the training data set. There are two types of learning problems in supervised learning.

  1. Regression problems — used to predict continuous valued output
  2. Classification problems — used to predict discrete valued output

Applications of Supervised Learning : Spam filters, Fraud detection, Handwritten character recognition

Supervised Learning — Image Source:Google

Unsupervised Learning

In unsupervised learning, we give the algorithm a set of data that doesn’t have any labels. We don’t tell what each data point is, and allowed to find some structure in the data by the algorithm. Unsupervised learning algorithm may break input data into separate clusters based on commonalities in the data and react based on the presence or absence of such commonalities in new input data. Cluster analysis or clustering is the most commonly used technique of unsupervised learning.

Applications of Unsupervised Learning : Organizing computing clusters, social network analysis, Market segmentation, Astronomical data analysis

Unsupervised Learning — Image source:Google

Reinforcement Learning

In reinforcement learning, a software agent learns without interference of humans by interacting with its environment. We don’t tell how to perform a task but the software agent works through the problem by itself. There is a method of reward and punishment to train the algorithm. The agent receives rewards by performing correctly and penalties for performing incorrectly. The agent practises to makes decisions to maximize its reward and minimize its penalty. Reinforcement learning is a best approach in artificial intelligence.

Applications of Reinforcement Learning : Autonomous vehicles, Games, robotics, deep learning

Reinforcement Learning — Image Source:Google

So, This is my simple introduction to Machine Learning. I think, you could get a simple idea about what is Machine Learning. Just go through the following to learn more about machine learning. It will be a great helpful to beginners in Machine Learning.

https://machinelearningmastery.com/

References:

https://www.youtube.com/playlist?list=PLLssT5z_DsK-h9vYZkQkYNWcItqhlRJLN

https://en.wikipedia.org/wiki/Machine_learning

See you next time.

Thank you !

--

--