ML World: Learn Machine Learning! Part 1

This tutorial answers all the following questions:

1. What is Machine Learning?
2. Why Machine Learning?
3. What is Labeled data and Unlabeled data?
4. What are the types of Machine Learning?

Congratulations on completing the Data Preprocessing Series. Welcome to ML world!

You have sorted out all the problems with dirty data, and now it is time we start our actual Machine Learning journey.

From here onwards, we will come across many ML models and implement various algorithms.

To begin with, we first need to understand what Machine learning is.

What is Machine learning?

I would put it this way: Machine Learning is teaching a computer program or an algorithm to improve on a particular task given to it. However, more practically, it is the study of how to build applications that exhibit this iterative improvement.

Many ways manifest this idea, but they are broadly recognized into three categories.

Why Machine Learning?

“Artificial Intelligence, deep learning, machine learning — whatever you’re doing if you don’t understand it — learn it. Because otherwise you’re going to be a dinosaur within 3 years.”- said Mark Cuban, a serial entrepreneur.

Machine learning has encompassed every possible field today!

It’s a huge deal: Machine Learning is the gush of the moment. Tons of companies are going all out to hire competent engineers, as ML is gradually becoming the brain behind intelligent systems. Through it, various fields are able to master consumers’ choices thereby increasing profits.

Today we have ML prevailing in the following sectors:

• Machine Learning in Data Analytics — CRM, Marketing Analytics, Audience Research.
• Machine learning for Predictive Analytics — Stock Market Forecasting, Market Research, Fraud Prevention.
• Service Personalization — Recommender Engines, User Modeling.
• Computer Vision — Image Recognition, Visual Search, Face Recognition
• Machine Learning Speech Recognition — Ai Assistants, Speech-To-Text, Automatic Subtitling

What are the types of Machine Learning?

The three major categories of ML are:

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

We live in a world that is instilled with Machine Learning and avid talks related to it. It will be interesting to identify these types as we encounter them.

For anyone looking to become a machine learning practitioner, it is of great importance to be able to identify the type of application to craft the algorithm appropriately.

For this, we first need to understand labeled data and unlabelled data.

Labeled data VS Unlabelled Data

What is Labelled Data?

The labeled dataset is one that has both input and output parameters.

In such datasets, both the training and the testing datasets are labeled.

Normally, if we are trying to predict a feature based on the others, that feature is the label. If we are trying to predict the weather we have (sunny or rainy), based on information on a particular day, then that is the label.

What is Unlabelled Data?

The dataset that has not been tagged with labels identifying characteristics, properties or classifications is known as Unlabeled dataset

Because the data does not have labels, the machine learning program has to identify each target based on its properties and characteristics.

One of the best ways to explain this is by using the example of pictures of different shapes.

Suppose the machine learning program is learning to identify three different Shapes–Square, Triangle, Circle.

If the data in the initial training dataset is labeled, the machine learning program works by matching successive images to one of those three categories.

If, however, none of the data is labeled with the three shape names: Square, Triangle, Circle, the machine learning program will need to work by evaluating each image and looking at characteristics like size, sides, angles, and other features.

What is Supervised Learning?

You can consider Supervised Learning analogous to a student learning in a school under the guidance of a teacher.

We have a labeled dataset that acts as a teacher, and its role is to train the model or algorithm.

Thus, Supervised Machine Learning is a type of predictive machine learning in which the data comes with Labels, where the label is the target we are interested in predicting.

Supervised Learning is often described as goal-oriented because of this.

It is highly focused on achieving a single goal, feeding more and more examples to the algorithm until it can accurately perform the required task.

Examples of Supervised Learning in Real-World:

• Speech Automated Systems: Cortana, Google Assistant, Siri, etc, or any other speech automated system on your mobile phone is trained on your voice and starts working based on this training.
• Spam Classification: If you use a modern email system, chances are you’ve encountered a spam filter. That spam filter is a supervised learning system.

What is Unsupervised Learning?

Unsupervised Learning is the opposite of Supervised Learning.

Here, our algorithm would be fed a lot of data and given the tools to understand the properties of the data.

From there, it can learn to group, cluster, organize the data in a way such that the newly organized data makes sense.

This makes Unsupervised Learning Data-Driven as it is based upon data and its properties.

The result of an unsupervised learning task is controlled by the data and the way it is formatted and organized.

Examples of Unsupervised Learning in Real-World:

• Recommendation System: Many times, after you browse for something you start receiving ads based on your search, this is one of the applications of supervised learning, where the machine learns your likes and suggests you increase chances of you clicking on it.
• Market segmentation: Dividing customers into groups based on demographic and purchasing behavior, to create different marketing strategies for the groups is another application of unsupervised learning.

Supervised Learning VS Unsupervised Learning:

What is Reinforcement Learning?

In this type of machine learning, an algorithm or machine is exposed to an environment where it gets trained by itself using trial and error methods.

In this case, the machine learns with experience and makes better decisions based on the feedback it received.

For reinforcement learning problems, we need an agent and an environment as well as a way to connect the two through a feedback loop.

To connect the agent to the environment, we give it a set of actions that affect the environment.

The agent is rewarded or penalized with a point for every right answer and wrong answer. In this way, the model gets trained to predict new data presented to it.

Examples of Reinforcement Learning in Real-World:

• Video Games: One of the most common places to look at reinforcement learning is in learning to play games. A computer learning to play chess by practicing moves to make the best decisions in any situation is an example of reinforcement learning.
• Robotics: For many robotic applications, it is useful to have our machines learn to complete their tasks without having to hardcode their processes.

Our world is drastically changing, machine learning is becoming increasingly dominant in everything we use each day.

Everyone, whether from a technological background or not should have a brief idea about Machine learning. It is the need of the hour!

Understanding even the fundamentals will help us to navigate this world, elucidating what seems like a lofty concept.

Happy Machine Learning!

(Image Source: Internet)