Machine Learning Problems

Machine Learning will be a property of every application.

Christopher Nguyen

These days when about 90% of data generated is unstructured, the reason for Big Data is Machine Learning with many problems to solve or to cope with in more efficient way.

Problems and examples

Example:

When your permissions allows to tag you on the photos, your friends probably will get some prompt from Facebook, which will help them to tag faster. In simply words Facebook helps people to recognize other people on theirs photo.

Example:

Self-driving cars are becoming very popular, but have you ever wondered how they are dealing with obstacles on the road? This is where machine learning comes and helps cars to avoid accidents and drive only on roads.

Example:

OCR programs transforms a manuscript, written character or scanned text document into digital version.

Example:

When you are playing on Xbox using Kinect you are definietly using (maybe sometimes without your knowledge) machine learning algorithms for gesture / posture recognition. Also swipe gestures on your mobile are taking advantages of these.

Example:

Have you noticed less unwanted emails in your inbox, say big thanks to data scientists who are working on it. They are probably grouping mails and / or searching for specific words in the text and then classify some messages as a spam.

Example:

Analyzing the patient symptoms and comparing them with a database of anonymized patient records can lead to computer driven prediction whether the patient is likely to have an illness.

Example:

Usually it is tough decision to sell or to buy current stock but after analysis past price movements, financial analysts will have an opportunity to make a better choice in this problem.

Example:

A lot of decisions these days are being taken on the opinion of others. We buy a product more because it has received a positive opinion and we visit a hotel most likely because it got the best rating online.

Example:

Extraction from text things like for example addresses, keywords, bank accounts, telephone numbers might be done using machine learning algorithms.

Example:

Find out if something found on the Internet is true or false is also very attractive for many people and companys, using machine learning we can investigate some opinions and statements to check if they are genuine.

Example:

What movie you should watch next? (Netflix), what song will make you happy? What product you should buy next? These questions are very common in huge amount of businesses these days.

Example:

At these days we something called information explosion for example - 50 Million Tweets per day, many of us have a need to group in some way these data, it is becomming more harder each day because of increasing data, so machine learning algorithms suits here pretty well.

Example:

Currently we are experiencing about 10,000 credit card transactions every second, this huge amount of data but how many of these are illegal? done by thiefs.

Example:

Mainly eCommerce and mCommerce have a need to better understand their clients by identifying users profile based on service usage. For example how many of their clients are Parents? How many of their clients lives in big cities?

Example:

This is one of the most common use case of machine learning. There is a big business value here and a lot of benefits for companies mainly in telecom and banking industries.

Example:

Siri, Cortana or any other speech assistant in your mobile was build mainly by data scientists on machine learning stack.

Example:

Reinforcement learning is very popular machine learning technique here, every game character who has to move within virtual environment probably was programmed using specific algorithm and is able to avoiding obstacles or enemies.

Example:

Predict the credit score of new customers in the bank or the next downtime of electricity, how much companies in these industries can save their money with such knowledge?

According to these problems and very popular startup movement I can fully agree with this statement:

Investors increasingly see data science (as a discipline) coming of age with more ways to solve real-world problems.

Prashant Sharma

Posted with : Machine Learning

If you liked this post, you can share it with your followers or follow me on Twitter!