What is Machine Learning ?

In this interesting world of A.I, Machine Learning will be a very common term that you will come across. Among other terminologies such as deep learning and neural networks.

Machine learning algorithms are ‘programs’ or ‘algorithms’ that find patterns in data and therefore get insight of the data enabling them to solve future problems by using their ‘experience’ and thus do not need to be explicitly programmed.

Simply, M.L Algorithms learn by analyzing hundreds and even thousands of examples and thus they can use what they have learnt from the data to solve future problems, problems that they have never encountered before.

There are 3 main ways in which machines can learn:

1. Supervised Learning : This is the case where you teach the machine (I mean the M.L Algorithm) by giving it numerous examples. In our world, this is the equivalent of a teacher showing the student examples of books and pencils. The student learns how the book differs from the pencil and thus can classify different types of books and pencils in future without the teacher’s help.
2. Unsupervised Learning : This is the case whereby you give the algorithm a bunch of data and let it figure things out on its own. In our world, this is equivalent to being thrown in the jungle and you have to figure your way out. For example by observing birds and insects or looking for paths that may lead you to freedom. Unsupervised algorithms are more difficult to implement as compared to Supervised Algorithms but they are more applicable in the real world and thus are highly valued.
3. Reinforcement Learning : This is an interesting way in which machines learn and correct themselves by being punished or rewarded for their deeds. In our world, this is equivalent to a parent punishing a child when wrong and rewarding the child when the child performs excellently.

As you may have observed, Machine Learning is modelled after human characteristics and this understanding is crucial in building your own machine learning models.

Machine Learning Models are simply the programs that we come up with after training the Algorithms. Yes, they are just programs, only that we give them fancy names (Machine Learning Models)

What about deep learning and neural nets? Deep Learning is a subset of Machine Learning. The major difference being, Deep Learning algorithms are modelled after the human brain and thus they work very similarly to the human brain. They are more efficient but they require large amounts of data to train and thus learn, as well as high computation power (which is expensive). Deep Learning is really interesting, but having a basic understanding of Machine Learning is required for you to fully realize its potential.

Remember Machine Learning models are simply names we give to our programs after training? Neural Nets (Simply) can be viewed as programs that are trained using the Deep Learning Approach. (More on this Later)

I hope you are enjoying this tutorial series so far.

For any Questions or Comments, Feel free to Contact Me.