Coping with AI errors with provable guarantees

Ivan Yu. Tyukin, Tatiana Tyukina, Daniël P. van Helden, Zedong Zheng, Evgeny M. Mirkes, Oliver J. Sutton, Qinghua Zhou, Alexander N. Gorban, Penelope M. Allison

Abstract: Highlights•In this work, we introduce a novel approach to address the challenge of random AI errors.•It enables correcting errors with provable guarantees regardless of data distribution and with small training sets.•The theory is illustrated with examples highlighting the application of the method to challenging machine learning problems.