University of Leicester
Browse
Separation_Notices6.pdf (464.43 kB)

Stochastic Separation Theorems: How Geometry May Help to Correct AI Errors

Download (464.43 kB)
Version 2 2023-08-03, 07:03
Version 1 2023-08-02, 16:02
journal contribution
posted on 2023-08-03, 07:03 authored by A Gorban, B Grechuk, I Tyukin

Recent years have seen explosive progress in data-driven artificial intelligence (AI) systems. Many decades of the development of mathematics underpinning statistical learning theory coupled with advancements in approximation theory, numerical analysis, technology, and computing gave rise to new-generation AI transforming our life. These systems show great promise in cancer diagnostics MSG$^{+}$20, they are a part of autonomous cars 22, automated face recognition and biometrics KE21, image segmentation SBKV$^{+}$20, language processing and translation tools DZS$^{+}$22, and as such become our new reality. Availability of unprecedented volumes of data, citizens’ expectations and participation are further driving this change.

History

Author affiliation

School of Computing and Mathematical Sciences, University of Leicester

Version

  • AM (Accepted Manuscript)

Published in

Notices of the American Mathematical Society

Volume

70

Issue

1

Pagination

25 - 33

Publisher

American Mathematical Society (AMS)

issn

0002-9920

eissn

1088-9477

Copyright date

2023

Available date

2023-08-03

Language

en

Usage metrics

    University of Leicester Publications

    Categories

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC