A Brief History of Machine Learning

When Was Machine Learning Invented?

It seems like machine learning is everywhere these days. Businesses increasingly rely on tools that use ML algorithms to provide them with accurate data for improving and growing their organization.

Many services we use daily, such as social media and Netflix, use ML to analyze consumer behavior and recommend trending content.

While machine learning may seem like a very recent concept, you may be surprised to know that the history of machine learning dates back to the 1940s.

However, it wasn’t until the 1950s that we saw how ML works for the first time.

Who Came up with Machine Learning and When?

We can’t say that a single person invented machine learning, especially the advanced ML algorithms we know today. Many genius individuals contributed to its development.

But there’s one person who stands out when thinking about when ML was invented.

Arthur Samuel, a computer scientist at IBM and a pioneer in AI and computer gaming, coined the term “Machine Learning” in 1952.

That was when he designed a computer program for playing checkers. The more the program played the game, the more it learned from its experience, thanks to a minimax algorithm for studying moves to come up with winning strategies.

Has Machine Learning Changed Much Since Then?

There have been numerous machine learning initiatives to date, helping ML evolve to a great extent since the ‘50s.

It didn’t take off until the late 1990s when IBM developed its Deep Blue supercomputer.

The showdown between Deep Blue and world chess champion Garry Kasparov was all everyone talked about back then. The chess computer beat Kasparov in 1997, proving that machines were indeed capable of human-like intelligence.

Since then, many scientists and researchers have jumped on the ML bandwagon and started developing new programs and algorithms.

However, they are all based on the first ML algorithms by Arthur Samuel.

We may have deep learning and AI-powered technology now, but those wouldn’t have been possible had it not been for machine learning.

The Evolution of Machine Learning

The history of machine learning is quite impressive. Let’s see where it all began and how it has evolved over the years.

1949 – Donald Hebb published “The Organization of Behavior,” introducing theories on the interaction between neurons, which were later crucial in developing machine learning.

1950 – Alan Turing invented the Turing Test, or the imitation game, to determine if a computer can pass for a human-based on its written linguistic fluency.

1951 – Dean Edmonds and Marvin Minsky built the SNARC machine, the first machine with an artificial neural network, based on Hebb’s model.

1952 – Arthur Samuel developed a computer game of checkers.

1957 – Frank Rosenblatt used Hebb’s model and Samuel’s ML algorithms to develop the perceptron, a computer program with human-like thought processes, primarily designed for image recognition.

1967 – Thomas Cover and Peter E. Hart came up with the nearest neighbor algorithm, which later became the foundation for pattern recognition. 

1979 – Stanford students built the Stanford Cart, a remotely-controlled, autonomous cart that could navigate on its own and avoid bumping into objects. It was designed to help study the remote control of a Moon rover.

1979 – Kunihiko Fukushima published a research paper on the Neocognitron, an artificial neural network (ANN) with multiple layers for detecting complex patterns. His work later inspired a convolution neural network (CNN) for deep learning.

1981 – Gerald DeJong proposed EBL (Explanation-Based Learning), a method that an ML algorithm can use when analyzing data to create general rules and ignore irrelevant data points.

1985 – Terrence Sejnowski invented NETtalk, an ML-based computer program that could perform cognitive tasks like a human. With written English text and phonetic transcriptions as input, it learned to “talk” like a baby. 

1986 – Paul Smolensky invented the restricted Boltzmann machine (RBM) for predicting probabilities of various possible outcomes based on input data. Today, this algorithm is commonly used for AI-driven recommendations and price predictions.

1990 – Robert Schapire introduced boosting algorithms for improving AI models. They consist of multiple weak classifiers that together create a strong learning model. Boosting algorithms are used today to analyze massive amounts of data and drive insights from the results.

1995 – Tin Kam Ho introduced a random forest algorithm that creates decision trees from multiple AI-powered predictions. Nowadays, it helps in driving accurate predictions from data and enhancing decision-making.

1997 – IBM’s supercomputer beat Garry Kasparov in chess.

2006 – Geoffrey Hinton invented fast-learning algorithms based on an RBM and came up with the term “Deep Learning” to explain how AI-based on ML can learn like a human.

2009 – Fei-Fei Li developed ImageNet, an image-based database for improving ML and AI, enabling them to learn from real-world data.

2011 – IBM’s Watson beats a human at Jeopardy, thanks to machine learning and natural language processing (NLP).

2012 – Google developed Google Brain, an ML algorithm for recognizing cats in images and videos. It marked a breakthrough in image processing.

2014 – Facebook created DeepFace, a facial recognition system capable of detecting faces in images, and accurately identifying humans.

2014 – Google introduced Sybil, an ML system for predicting user behavior, designed primarily for better advertising.

2014 – Chatbot Eugene Goostman passed the Turing Test, although it convinced only 33% of the competition judges that it was human.

2016 – Google’s AI-powered AlphaGo beat a professional player in Go, an abstract strategy board game.

Final Thoughts on the History of Machine Learning

These days, machine learning has numerous applications and can help every business become a data-driven organization.

It has become an integral part of various technologies, as ML algorithms can detect fraud, predict prices and stock market trends, analyze and streamline sales data, personalize user experience, recommend products, power up autonomous cars, and much more.

The future of machine learning shines bright, and we can’t wait to see what it has in store for us next. It’s undoubtedly going to be brilliant and exceed our wildest expectations.