A method that allows a computer to give advice and to "train" another computer in a way that refers to "teacher" and "student" developed researchers School of Electrical Engineering and Computer Science of Washington State University.
The paper was published online in Connection Science, and led by researchs was Matthew E. Taylor, Professor of Artificial Intelligence. As it became known, the researchers put their "agents" (as the virtual robots used in the study are characterized) to act as teacher-student pairs: the goal was to train the "students" in two electronic games, Pac- Man and a version of Starcraft. As the research showed, the "student" was able to learn the games, even surpassing the teacher.
Many sci-fi fans' minds will go straight to menacing digital figures such as 'Skynet' from 'Terminator' or the 'Cylons' from 'Battlestar' Galactica”, however, according to Taylor, robots are not going to take over the world in the near future, as “they are too dumb”. As he points out, even the most sophisticated robots get confused easily – and when they do, they stop working. Very often it takes two to three times longer than he thinks is required to make a robot work, he adds.
Computer literacy in computer games is an important part of robotics research. In this way, robots will be able to train each other in new jobs without the need for human intervention: for example, a cleaning robot will be able to train his replacement.
According to Taylor, the best way to properly train a robot in new activities is to transfer the "brain" of an "experienced" predecessor to it. However, there are problems when hardware and software are incompatible with the new model.
In the context of the study, program researcherseyelike "teachers" focusing on advice over action - specifically, when the "teacher" tells the "student" to act. "We've designed algorithms to give advice, and we're trying to figure out when our advice has the most impact," says Taylor.
Source: naftemporiki.gr
