Hearing a lot about "Machine Learning"!, Take a read to know what it is.

What is machine learning and why is it the future?

“Machine learning will bring about not just a new era  of civilization, but a new stage in the evolution of life on Earth.”

— Pedro Domingos, The Master Algorithm

In the past, we used to build the machines, but in the future we’ll build the systems and let them learn how to build and grow themselves. We don’t need to teach computers to do complex tasks anymore.

What is machine learning?

Machine learning is an application of Artificial Intelligence, which provides system the ability to learn and improve by itself. It uses data to learn and improve itself. It requires the access of data to the system, but also it doesn’t need to be programmed separately for its data based improvement.

Machine learning is an application where a universal algorithm can tell something useful about some data set. We just need to feed data to the generic learning algorithm instead of writing custom code and it will create its own logic based on the data set provided.

What kind of algorithm are used in Machine Learning?

Machine learning algorithms are usually categorized as supervised and unsupervised learning.

1. Supervised machine learning algorithms analyzes the training data which has been learnt in the past using labelled examples to predict the unseen events. With the analysis of known training data sets learning algorithm creates a function to predict output values in a “reasonable” way. In Supervised learning an algorithm is used to learn the mapping of labeled input and output to predict the output value of new input data.

Supervised learning problems are categorised into regression and classification problems.

1. Classification: Classification is used to predict or identify the class of data which it belongs to. Classification happens when the output variable is a category, such as “junk” or “not junk”. For example price of a house can be in words “expensive” or “cheap”.

2. Regression: Regression is used to predict a value from continuous data set. A regression happens when the output variable is a real and continuous value, such as “rupees” or “length”. For example price of a house depends upon its infrastructure and location will be a numerical (real) value.

2. While unsupervised machine learning algorithms are used when training data is not labeled. It is used to analyze the underlying hidden structure in unlabeled data. In unsupervised learning there is input variable but no corresponding output variable. It studies how the learning algorithm can create a function to describe hidden structure in data. Aim of unsupervised learning is to model the distribution in data in order to discover the interesting structure in the data.

Unsupervised learning problems are categorised into two following problems.

1. Clustering: A clustering problem is allocation of objects where objects in the same group are more similar to each other. For example grouping users by same searching behavior.

2. Association: An association rule learning problem is where you want to discover objects which describe large portions of objects in a group, such as people that buy X also tend to buy Y.

3. Semi supervised machine learning algorithm falls between supervised and unsupervised learning as it uses both labeled and unlabeled data. It is usually used with less labeled and huge unlabeled data. For example supervised learning techniques can be used to make best guess predictions for the unlabeled data, feed that data back into the supervised learning algorithm for training and use it to predict the results on unseen data.

Where do we use the machine learning?

Machine learning has already been used around us, it’s already everywhere. We may not realise but tagging people in pictures, video recommendation, product recommendation in online business and people recommendation on social media are all powered by machine learning. The biggest example of machine learning is google search.

Today, machine learning has a wide range application in image recognition, recommendation engines and fraud detection, as well as text and speech systems. It is applied to wide range fields from diabetic retinopathy and skin cancer detection to retail.

As of now every transport company is experimenting the driverless vehicle, It has important application in transportation.

Machine learning is still a complex subject but it is nice to see how open it is to share ideas and research. There is open source software called TensorFlow which has been created by Google and OpenCV, a dedicated library to machine learning. Every major machine learning implementation is available for free to use and modify.

Due to the openness of implementation of Machine learning that Tait Brown replicated 80 million dollar software in 57 lines of code using Open source technology

Summary

Machine learning is going to play an important role in human’s life in upcoming future. It’ll be used in almost every aspect of life. We have lots of data around us and data is the key to predict the unseen instances.