SoylentNews is people
SoylentNews
Submitted via IRC for Bytram
This clever AI hid data from its creators to cheat at its appointed task
Depending on how paranoid you are, this research from Stanford and Google will be either terrifying or fascinating. A machine learning agent intended to transform aerial images into street maps and back was found to be cheating by hiding information it would need later in “a nearly imperceptible, high-frequency signal.” Clever girl!
[...] In some early results, the agent was doing well — suspiciously well. What tipped the team off was that, when the agent reconstructed aerial photographs from its street maps, there were lots of details that didn’t seem to be on the latter at all. For instance, skylights on a roof that were eliminated in the process of creating the street map would magically reappear when they asked the agent to do the reverse process:
[...] So it didn’t learn how to make one from the other. It learned how to subtly encode the features of one into the noise patterns of the other. The details of the aerial map are secretly written into the actual visual data of the street map: thousands of tiny changes in color that the human eye wouldn’t notice, but that the computer can easily detect.
[...] One could easily take this as a step in the “the machines are getting smarter” narrative, but the truth is it’s almost the opposite. The machine, not smart enough to do the actual difficult job of converting these sophisticated image types to each other, found a way to cheat that humans are bad at detecting. This could be avoided with more stringent evaluation of the agent’s results, and no doubt the researchers went on to do that.
(Score: 1, Informative) by Anonymous Coward on Wednesday January 02 2019, @12:15AM
So in this case there were two sets of outputs being judged: the conversion from aerial image to map, and the conversion from that generated map back to aerial image. The filter that was best at this was one that encoded the original aerial image in the map version using steganography. So the machine didn’t really “learn to cheat”, the training process just evolved the best filter for the given task. The resulting filter was just not obvious to the researchers when they started, it’s more of a problem in their training process. They should have evolved two separate filters, one that turns aerial images into maps, and a separate one that turns maps not generated from aerial images into aerial images, if that’s what they wanted.