Lee Bressler discusses the history of machine learning technology

Today’s machine learning technology allows computers to communicate readily with humans, diagnose serious medical conditions, and autonomously drive cars and trucks, among countless other applications says Lee Bressler, an investor and technologist from New York with a focus on artificial intelligence and machine learning technology. While these advances are, for the most part, at the forefront of modern technology, machine learning has its roots in the 1950s, and—by some definitions—many, many years before that.

Bressler goes on to explain that, in 1952, another machine learning pioneer, Arthur Samuel, developed the first-ever computer learning program. Studying the game of checkers, it successfully improved its strategy—without outside interference—by repeatedly playing and analyzing which moves provided the best chance of winning, before incorporating them into its program.

“The first neural network for computers came five years later,” Bressler further reveals, “designed by Frank Rosenblatt to simulate the thought processes of the human brain.”

The 1960s, 1970s, and 1980s saw continued advances in machine learning. “Over the course of the three decades that followed, pattern recognition and map routing algorithms were developed, as were robotic devices capable of navigating obstacles,” explains Lee Bressler. The same period also saw the first examples of explanation-based learning – a form of machine learning capable of exploiting domain theory in order to form concepts from training examples.

Approaches to machine learning in the 1990s switched from knowledge-driven to data-driven, with algorithms designed to analyze vast quantities of data before drawing conclusions and effectively learning from the results. “1997 saw another huge leap forward for machine learning when IBM’s ‘Deep Blue’ chess-playing computer successfully beat the then-world champion at the game,” adds Bressler.

In the 21 years since, Geoffrey Hinton coined the term ‘deep learning’ wherein which computers are able to ‘see’ rather than simply processing data. Along the same lines, Microsoft developed never-before-seen motion tracking software, capable of reading and reacting to human movement and gestures, back in 2010. Meanwhile, in 2011, another IBM creation—this time called ‘Watson’—successfully beat a team of human competitors at Jeopardy, further demonstrating the possibilities afforded by machine learning technology.

“In the last year or two alone, we’ve seen huge further advances, including in 2016 when Google’s AlphaGo artificial intelligence algorithm successfully beat a professional player at what is considered the world’s most complex board game, ‘Go,'” adds Lee Bressler, in closing, “and I personally cannot wait to see what else the future of machine learning technology holds.”