One day Mickey was going home when he saw Minnie on the Brooklyn bridge. He ran to her.
Mickey – Hey Minnie, How are you?
Minnie – I am okay!
Mickey – What happened, what are you doing here.
Minnie – Just came out for a walk on the bridge. I have been in my home all day trying to understand what exactly is Machine Learning. Looks like I haven’t got it.
Mickey – No worries, I know you understand better with a Story.
Minnie – What! a story on Machine Learning?… Wow
Mickey – Yeah, we first need to understand Why exactly do we want Machines to Learn.
Minnie – Okay
Mickey – Suppose you need to go for a holiday to Paris. Now you have to estimate the amount you will need to save for this holiday. How much amount will you save monthly?
Minnie – Well it depends. I will have to do some analysis on this.
Mickey – What analysis will you do?
Minnie – For example, I will see what will be the flight tickets at that time. How much money will I need to carry? It will depend on the season of the year, hotel rates etc.
Mickey – Yes, And from where will you get all this information
Minnie – Well, I will get the data from different websites, or can also get some inputs from friends, online, offline, my past holiday experiences, etc.
Mickey – Exactly, so you predict the price of your holiday based on some known historical data. This is called Regression in Machine Learning i.e. predicting based on data
Minnie – This is getting interesting.
Mickey – And we do this all the time when you have to buy a car, a home, or even prepare your monthly grocery budget.
Minnie – Yeah
Mickey – Now for an average person doing all the calculations based on historical data may not be very easy and therefore we need machines to do the math.
Minnie – So is my calculator also based on Machine Learning
Mickey – No, Calculator works on a set pattern. See here we provide machines as much data as possible so that it can find patterns and predict the outcome. Like the cost of your holiday.
The calculator can just do mathematical calculations for you like ‘+’ ‘-‘, but cannot predict the cost of your holiday.
Minnie – I am getting it.
Mickey – Let me show you the Wikipedia definition of Machine Learning (ML)
“Machine learning is the study of computer algorithms that improve automatically through experience. Machine learning algorithms build a mathematical model based on sample data, known as “training data“, in order to make predictions or decisions without being explicitly programmed to do so.”
Minnie – This now looks so much understandable.
Mickey – I know
Minnie – So is that all about ML
Mickey – No, now with this understanding let us see the components of machine learning
Minnie – Wow, this is getting exciting
Mickey – So we know Machine Learning depends of the data we feed to the machine (i.e. computer, robot, etc)
The more and greater variety of data we feed, the better it can create patterns and predict the results.
For example, if you mark some emails as Spam in your inbox, over time the system will understand your preferences based on the data you have provided and in future will automatically mark the emails as spam if they match the predictions.
Minnie – I can now relate to it.
Mickey – So we know Data is important in ML
Now there are 2 main ways to get data – manual and automatic. This collection of data is called dataset
Minnie – Noted
Mickey – The 2nd important component of ML is Features.
Feature is an individual measurable property or characteristic. For example in the case of calculating the cost of holiday features can be flight services, no of days, route, etc. Features can be numerical, string or even graphical.
Minnie – Great, I got Dataset and Features
Mickey – The 3rd component of ML is Algorithm. It is a sequence of instructions to solve a problem
Minnie – Can you give some easy real-world example that I can relate to
Mickey – Sure, have you ever wondered when you work online how come you get recommendations and ads matching closely to your interests?
Minnie – Yeah, last week I searched for a holiday package to Italy and now I keep on getting recommendations on that.
Mickey – Exactly, the system has taken your inputs and based on the data and features, it is showing you all adds as processed by the algorithms of ML.
Minnie – It all sounds so nice now
Mickey – I know
Minnie – One more question. Where do then Artificial Intelligence and Deep learning fit in the picture?
Mickey – Well you can say ML is a subset of Artificial Intelligence (AI) and Deep Learning (DL) is the subset of machine learning
Minnie – Oh I see.
Mickey – But more on that later.
A Story of Artificial Intelligence
Minnie – I now can understand ML
Mickey – There are 3 types OR approaches of ML
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
Supervised Learning:
Finds patterns using both, input data and output data.
example:
- predict and find spam emails
- Predict the winner or loser of a game
- The decision to give a Loan or not
Unsupervised Learning:
Find patterns based only on input data. This technique is useful when you’re not quite sure what to look for.
example:
- Visual Recognition – Like recognizing your friends in a photo on Facebook
- Google Photos – creating albums on friends based on facial features.
Reinforcement Learning:
A computer program interacts with a dynamic environment in which it must perform a certain goal (such as driving a vehicle or playing a game against an opponent)
example
- Automated Traffic Lights – to solve the congestion problem
Minnie – Okay, I can understand now.
Mickey – Great, do you want to go home now
Minnie – Well, google map shows traffic congestion and predicts 1hr to reach home. Let’s wait.
Mickey – So how does the map predicted time to reach your home.
Minnie – Wait… That’s Machine Learning?
Mickey – You are getting smarter.
Minnie – Thanks to you.
Mickey – I am always here for you 🙂
