Scientists Teach Machines to Learn Like Humans — Artificial Intelligence Advances

They then compared the outputs from both humans and machines through ‘visual Turing tests. A robot that can make logical leaps about things might someday be more adept at human-like decision-making. The software is not mimicking the way children acquire the ability to read and write but, rather, the way adults, who already know how, learn to recognize and re-create new characters.

The study authors stated further that considering the low level of data needed by humans to learn, the present effort by the researchers could lead to new advances in the field of artificial intelligence.

Brenden Lake, the study’s lead author and a Moore-Sloan Data Science Fellow at New York University, said the algorithm can lead to the development of better learning machines.

Here, the researchers’ model specified that characters in human writing systems consist of strokes, demarcated by the lifting of the pen, and that the strokes consist of substrokes, demarcated by points at which the pen’s velocity is zero.

After learning the characters, the computer and humans were asked to copy several handwritten characters.

Only about less than a quarter of the judges were able to assess the handwritten characters correctly if they have been reproduced by a machine or a human.

The Washington Post reports that the algorithm helped computers to recognize letters from a limited set of examples. Google’s search machine learns your behavior and proceed to give you results accordingly. It was done after both were shown just one example of each character. “I think you can have the best of both worlds”, he says.

Just like the current machines that learn from examples, Chappie also learns from the people that teach him. Now imagine somebody shows you a pineapple.

Moreover, each program is created to generate variations of each symbol whenever the programs are run, helping it capture the way instances of such concepts might vary, such as the differences between how two people draw a letter.

Human volunteers from Amazon’s Mechanical Turk service were recruited to hand-draw thousands of characters from 50 different alphabets, including traditional languages like Latin, Greek, and Korean, as well as fictional ones like the alien language in “Futurama”.

Human judges were then asked to identify which characters were reproduced by a computer. But, when machines are allowed to perform new tasks they can only replicate those performed earlier by humans. But the research could also shed fresh light on how learning happens in young humans, the scientists pointed out. They don’t see characters as just inactive objects that are visual. The new AI averaged 3.3% errors, whereas competitor programs varied between 8% and 34.7%. If the interrogator couldn’t reliably distinguish the human answers from the computer answers, the computer would pass the test and have the status of a thinking machine, Turing argued.

Computers could reach conclusions after sifting through hundreds of thousands of examples, but scientists have had a hard time getting them to make inferences and comparisons during machine learning.

AI is still a long way from matching human abilities. The findings claim that computers have come closer to the human method of learning, The New York Times reported.

But our ability to quickly extrapolate from a small set of data comes with some notable drawbacks.

“I would be very careful with terms like ‘superhuman performance, ‘” said Oren Etzioni, chief executive of the Allen Institute for Artificial Intelligence in Seattle. “It’s the inevitable flip side of learning so quickly”.

The Turing test was first proposed by British scientist Alan Turing in the 1940s as a way to test whether the product of an artificial intelligence or computer program can fool humans it’s been made by humans. And while this research is important, it doesn’t necessarily mean that this is the way all machines will learn in the future.

The study, published in the journal Science, describes how scientists developed a “Bayesian Program Learning”, or BPL algorithm.

Each step, each paper, and each idea lights another candle to illuminate the massive void in our knowledge of intelligence and consciousness. Maybe tomorrow it will generating human-like speech, or even recreating art with greater success.