← All Posts /


Machine learning, as the name sounds, means “educating” a machine to behave, adapt and perform on its own. Machine learning focuses on building a computer program, that can learn and adapt to new data without human interference.

Machine Learning for the future For instance, here are two equations- x + y = 10 and x = 5.

Now, there are many ways to reach the solution, y= 5. This process of creating an algorithm through which the machine reaches the solution is known as Machine Learning.

Machine learning algorithms can be broadly categorized into –

1) Supervised learning

This describes creating a function f(x) which maps the inputs to outputs.

To explain in detail, let’s say an infant is shown three different types of fruits by his teacher – an apple, a banana, and an orange. The teacher explains how to differentiate between these fruits, using their features. So now, the infant knows that a yellow-coloured elongated fruit is called a banana, and similarly, he knows what an orange or an apple looks like. The teacher supervised the learning of the infant in this case. This example is representative of how supervised learning takes place.

There are 2 main types of supervised learning problems-

Classification- It involves predicting class labels. Regression- It involves predicting numerical labels. 2) Unsupervised learning

This describes creating a model which extracts and devises relationship within the data.

To explain in detail, let’s say the same infant is shown the same threefruits – an apple, a banana, and an orange. However, no teacher is present this time to supervise him. The infant, even if he doesn’t know the names of these fruits, he can see that all of these fruits have different features, that is, different shapes and colours, and so he will learn to differentiate between these fruits. The next time the infant sees a fruit that is spherical and orange-coloured, he’ll know that it is different from red-coloured or yellow-coloured fruits.This example is representative of how unsupervised learning takes place.

There are 2 main types of unsupervised learning problems-

Clustering- It involves grouping within the data Density Estimation- It involves summarizing the distribution of data APPLICATIONS OF MACHINE LEARNING

Machine learning is widely applicable in this dynamic world. It aims at building a model through complex algorithms. The model uses the codes and algorithms to form various patterns for the decision-making process.

In this “programmed” and “online” world, machine learning can be used by-

1) Trading firms to find good performing assets or equities by finding new patterns in data using techniques such as clustering.

2) E-commerce and marketing firms to create a recommender system, which helps provide accurate and personalized recommendations to users based on their past orders or search history.

3) Banks and lending institutions to build credit risk models and find credit rating for their users. Frauds and faulty transactions can be identified using anomaly detection algorithms.

These examples are just indicative; the actual applications of machine learning are endless. We intend to cover much more of this topic in the upcoming posts. So, until then, stay tuned!

Recommended » Benford's Law