Introduction to Machine Learning
Introduction to Machine Learning
Machine learning is one of the hottest buzzwords in this 21st century and our world is driving towards it. The below article is for the audience who wants to know about the fundamentals of machine learning.
The following are the topics I am going to discuss here
1. What is Machine learning?
2. History of Machine Learning
3. Types of Machine learning
1. What is Machine Learning?
We all know that human beings learn from experiences and machines follow the instructions to perform certain tasks. But what if machines can be trained to learn from past experiences (or we can say data) and perform their task whenever required? Yes, that is possible and that’s where the concepts of Machine Learning come into the picture. Let me explain this in layman’s words.
Suppose, a kid is shown, different pictures of a cat and dog to make him understand what a cat and a dog look like. He identifies some unique features of cats and dogs. The next time when he sees a neighbor’s pet, he recognizes it as a cat, not a dog. Here from his past experiences, he was able to identify the features of a cat in a neighbor’s pet and classify it as a cat.
The above picture represents how a kid is trained with different pictures of cats and dogs and is able to identify the neighbor’s pet as a cat.
Similarly, in machine learning, machines try to do the same thing that the kid did. It tries to recognize the patterns in the data through training and is able to predict new data.
As per formal definition, Machine learning can be defined as a subset of Artificial Intelligence (AI) that facilitates computer systems to learn from data independently using algorithms and can learn from an abundant amount of data. It can also be called predictive analytics as it helps to predict or classify in the context of new data.
In the above image, a set of training data is given as input to the ML algorithm(machines), it gets trained by itself and builds a model (classifier). When new data known as test data is fed to the model, it evaluates the results.
History of Machine Learning
Though machine learning has its popularity in recent decades, this is not new as its roots originate from dates back to 1952 when Arthur Samuel first came up with the phrase “Machine Learning”. He developed a game that could learn as it runs. Later on, in 1958, Frank Rosenblatt designed the first artificial neural network. Since then there has been a lot of advancement in this field and we all know how it became an integral part of life in the 21st century. To know more about the history of ML please visit the link below:
https://labelyourdata.com/articles/history-of-machine-learning-how-did-it-all-start/
Types of Machine Learning
Machine learning is broadly classified into three types and we will be discussing each one of them.
1. Supervised Learning
2. Unsupervised learning
3. Reinforcement learning
The following figure represents the classification of machine learning
1. Supervised Machine Learning
This type of machine learning needs supervision by a teacher and is thus called supervised ML. This is a very popular ML method where a machine is fed with labeled datasets which act as a teacher for the machines. The machine gets trained with the data and whenever new data is given to it can make predictions over the new data.
This is again two types:
a.) Regression: In this type of learning, the output has continuous value to predict by machines. The following dataset in the fig below is an example of simple linear regression, where:
Input data: Years of experience
Output data (labeled): Salary (which is a continuous value).
Here regression algorithms will be trained with this dataset to predict Salary based on the value of Years of experience.
b.) Classification: In this type of learning, the output is having defined labels (having categories) or discrete values. The following dataset is an example of Classification where:
Input data: Attr1, Attr2, Attr3
Output data (labeled): Target Label (which is categorical).
Here classification algorithms will be trained with this dataset to predict Target Label (A or B) based on the other Input variable.
2. Un-Supervised Learning:
This type of learning guides the model to understand the structure of data and to identify patterns and relationships in the data. The dataset fed to the algorithm is unlabeled and it cannot predict any output.
The following figure is an example of unsupervised Machine learning
In the above figure, the ML algorithm is trained with a dataset having a combination of apples, grapes, and chocolates. The model tries to understand similar types of data and divide the data into different groups having similar data namely apples, grapes, and chocolates. This is an example of clustering.
Besides this, there are other unsupervised learnings like Dimensionality Reduction, Market Basket Analysis, Recommendation System, etc.
3. Reinforcement Learning:
In this type of learning a machine/agent interacts with the environment and takes actions based on the environmental condition. It works on trial and error methodology. The agent takes certain actions in an environment, if the output of the action is correct, it will be rewarded and if the output is wrong it will be penalized. The agent takes actions again and again until it gets the correct output. The most common example of reinforcement learning is the Driverless Car.
Conclusion
In this post, I have discussed the Basics of ML, the History of ML, and the Types of Machine learning briefly. There are a lot more things to learn in Machine Learning if you want to dig deeper into it.
Excited to learn Machine Learning? Open to receiving your suggestions and feedback. You can post your questions in the comment section and I will try my best to answer your questions.
