
The machine build by the MIT crew is now capable of fluently writting in Sanskrit, Gujarati, Glagolitic and Tibetan.
In the era of technological marvels, it bears no lasting impression to find out that research into intelligent machines is being done. But what if we heard about a machine actually capable of mimicking our way of thinking? Researchers built a machine which thinks like a child, machine capable of taking the actual learning process to a whole new level.
This research endeavor has been conducted by a team of researchers from the Massachusetts Institute of technology. The team of scientists was led by the esteemed professor Joshua Tenenbaum, an expert in the field of artificial intelligence. Tenenbaum stated that he and his team was able to devise a system which is capable of learning an entirely new concept or concepts just by analyzing a few examples.
The professor said that the team’s efforts were to imbue the machine with the learning skills of a child, who’s brains are capable of absorbing a high amount of information from different sources. Although the machine’s capabilities are highly limited, the team managed to pull a few aces out of their sleeves. Thus, the machine was able to learn how handwriting works by closing analyzing several characters from the alphabet.
Researchers built a machine which thinks like a child in order to develop the foundation for various application which require the use of an artificial intelligence. The learning abilities of a child were not arbitrarily chosen. As Tenenbaum himself points out, even before a child goes to kindergarten or school, he is highly capable of extrapolating and understanding various concepts just by looking a some examples. Moreover, the brain is capable of going over that, by providing new examples to the learned concepts.
By teaching it to recognize handwritten alphabet characters, the scientists have taken ample steps into learning how to feed the machine additional information concerning other forms of expression. For example, the machine’s computational capabilities could be further expanded if the scientists were to provide an example from other forms of expression such as dancing or singing. And, even more, intriguing is that, in the future, the machine could be introduced in some lessons regarding visual concepts.
The whole system was named the Bayesian Program learning, and its made up of several learning programs, integrated into the same network. Tenenbaum declared that the team is far from perfectly mimicking the learning capability of a young child, but important steps have been taken into that direction.
Researchers build a machine which thinks like a child and already the gizmo learned approximately 1600 examples of handwritten character, consistent with various writing systems. Basically, the machine is now capable of expressing itself in writing using Sanskrit, Gujarati, Glagolitic and Tibetan writing systems.
Photo credits:www.wikipedia.org