History of Machine Learning

History of Machine Learning- Every Techie Should Know!

“Netflix saved $1 billion this year as a result of its machine learning algorithm which recommends personalized TV shows and movies to subscribers.” “Amazon’s current ML algorithm has decreased the ‘click-to-ship’ time by 225%.” The two statements presented above clearly indicate to the increasing application of machine learning today and its direct business impact. In fact, machine learning features among the top three in demand skills, according to Monster.com. While there is no spec of doubt that machine learning is redefining the future, we believe it is a good idea to have a brief look at the history of machine learning. As techies and developers interacting with machine learning on a daily basis, exploring the history of machine learning, its origins and milestones will do you no harm. Who knows, it might just motivate you to delve deeper into the discipline. 

Timeline of Machine Learning

 

The Period of Research (Pre 1940s)

Way before my time and yours were a bunch of thinkers and inventors who had this ‘crazy nut’ in their brains which led to computing machines. Right from the  building of the first mechanical calculator to AI hitting the silver screen, each milestone laid the groundwork for what we see as machine learning today. This period of research in the pre 1940 days envisaged machine learning and AI as simply science fiction. Little did the researchers know that what they see as fiction then would become reality in a couple of decades. History of machine learning began with extensive research without an actual vision of translating into what we understand as machine learning today.  

Period of Research

From Intent to Action (Till 1980)

It was precisely in 1943 with the first neural network that seeds for turning fiction into reality were sown. Following that research and application took pace to the speed of light in the history of machine learning. The turing test for a computer to fool a human into believing the former was a human too became a proof of real intelligence. Then came the first computer learning program by IBM, playing checkers and improving along the game. 1957 was the year of Perceptron, the first artificial neural network, capable of pattern and shape recognition.

The nearest neighbour algorithm was born in 1967, laying the framework for very basic pattern recognition. More research, greater the application is something we are all familiar with. 1979 was the year when efforts of students from Stanford University bore fruit. Their ‘Stanford Cart’ came with the ability to navigate obstacles. In fact, it navigated a chair filled room without human help. Until then, such scenarios were reserved for the big screen, but machine learning brought them to real life.

Translatring Research to Application

Science Fiction to Reality (1980-2000)

Steady incremental developments in history of machine learning started bringing to life some of the most far fetched science fiction. In 1981, the world was introduced to explanation based learning, allowing the computer to discard unimportant data. 1985 is an important milestone to remember. While until now the focus was on data and action, NetTalk was an exciting entry. Interestingly, it had the ability to learn how to pronounce words, just like a baby does. It was another step towards replicating human intelligence development in machines. 

IBM’s Deep Blue stole the show in 1997 by beating the human world chess champion. 1999 saw computer and machines detecting cancer more effectively. While not being able to cure cancer, the CAD Prototype Intelligent Workstation detected cancer 52% more accurately than radiologists did. The period of 1980-2000 without a doubt translated the previous groundwork into reality and laid foundation for modern machine learning that is directly delivering business benefits

Computer beating human at chess

 

Modern Machine Learning (Post 2000)

2006 came as the year when the term ‘Deep Learning’ was coined. The objective was to elucidate that computers could see and differentiate objects and texts in images and videos. Voice recognition and image tagging are a modern day application of deep learning. “IBM’s Watson beats human competitors in Jeopardy” made headlines in 2011. The same year saw the inception of Google Brain capable of discovering and categorizing objects like a cat. The Google neural network could detect faces with 81.7% accuracy. 2014 was the year when Facebook developed DeepFace. Interestingly, it had the ability to recognize individuals on photos just like humans. Don’t you frequently get notifications that someone uploaded a photo that might include you, well, now you know how Facebook recognizes that. 

With Netflix, FB and Google in the lead, Amazon was not far behind by developing its own machine learning algorithm in 2015. In 2016, Google’s AI algorithm beats the best human competitor playing the hardest board game known to mankind. The modern era of machine learning is spurring a lot of meaningful collaborations. Humans and machines are siding together to beat online fraud, trolling and creating a better quality of life. 

Modern Machine Learning

From History of Machine Learning to its Future

What technology has in store for us will only unfold with the new day. But, one thing is for sure, machine learning and its collaboration with other digital disruptions like IoT, Virtual reality are definitely setting the tone for a more meaningful and productive future. If you wish to power your business growth with machine learning, get in touch with Recro today and leverage from our wide range of machine learning services.

 

Face Recognition

About RECRO

At Recro we combine Design Thinking, Lean and Agile to help you achieve market-leading  performance.

We partner with our clients with the aim to transform their business challenges into growth opportunities.

Pages

Categories

Subscribe



7869