ML Series 1: When computer was still a human being!
The aim of this series is to explain the corpus of machine learning, from the beginning to the present, in simple language accessible to everyone.
Shall we?
In earlier times, humans relied on their physical senses to solve problems and understand the environment. Counting has been documented since about 30,000 BC while ancient civilizations in China, Babylonia, India and Persia used mathematical models to improve the daily lives of their people. However, the sheer amount of computational power needed to resolve the models limited their use to simple cases.
And then came the dawn of the 20th century.
At that time, most calculations were done by humans. Mechanical tools such as Abacus, Napier’s Bones and Slide Rule were used to solve digital calculations. They were referred to as calculating machines or calculators, and the human who operated these machines were called the computers. Computers used to be humans!
The need for building advanced models and solving complex calculations during World War 2 led to advancements in calculating machines and unprecedented growth in their computational powers. Colossus, the world’s first electronic computer, was built by the British to decipher encrypted messages between Hitler and his generals. The Americans later built ENIAC which was the most powerful calculating device of the time. It was the first programmable general-purpose electronic digital computer.
As ENIAC was started in 1945, it became possible for the first time to use mathematical modeling for solving practical problems of significant size. The improvements in computer technology in the years since and the enormous gain in storage capacity and speed have made mathematical modeling increasingly attractive for civilian and military uses.
Launched 1951, the UNIVAC I mainframe computer was used to predict the outcome of the 1952 U.S. presidential election. The computer predicted a landslide for Dwight D. Eisenhower, whereas the final Gallup poll predicted a Eisenhower win in a close contest. It turned out that the computer was right and the humans were wrong.
The United States Army requested for the purchase of a UNIVAC computer. In the words of Colonel Wade Heavey, “This is a tremendous calculating process…there are equations that cannot be solved by hand or by electrically operated computing machines because they involve millions of relationships that would take a lifetime to figure out.”
The final act in the advancement of computer was the development of magnetic core memory. In 1949, Massachusetts Institute of Technology began to investigate the idea of using a magnetic core to replace the usual drums used in earlier models of computer. The Memory Test Computer was completed in 1953, and the computer deployed in 1954 for the first simulations of neural networks.
With this feat, magnetic core soon replaced other forms of high-speed memory.
Next, we discuss how machines became smarter with their newly discovered superpowers.
