The neural networks are harder to fool thanks to adversarial training. Los Alamos National Laboratory researchers have developed a novel method for comparing neural networks that looks into the "black box" of artificial intelligence to help researchers compre…