Počet záznamů: 1
Robust Multilayer Perceptrons: Robust Loss Functions and Their Derivatives
- 1.
SYSNO ASEP 0524790 Druh ASEP C - Konferenční příspěvek (mezinárodní konf.) Zařazení RIV D - Článek ve sborníku Název Robust Multilayer Perceptrons: Robust Loss Functions and Their Derivatives Tvůrce(i) Kalina, Jan (UIVT-O) RID, SAI, ORCID
Vidnerová, Petra (UIVT-O) RID, SAI, ORCIDZdroj.dok. Proceedings of the 21st EANN (Engineering Applications of Neural Networks) 2020 Conference. - Cham : Springer, 2020 / Iliadis L. ; Parvanov Angelov P. ; Jayne C. ; Pimenidis E. - ISSN 2661-8141 - ISBN 978-3-030-48790-4 Rozsah stran s. 546-557 Poč.str. 12 s. Forma vydání Tištěná - P Akce EANN 2020: International Conference on Engineering Applications of Neural Networks /21./ Datum konání 05.06.2020 - 07.06.2020 Místo konání Halkidiki Země GR - Řecko Typ akce WRD Jazyk dok. eng - angličtina Země vyd. CH - Švýcarsko Klíč. slova Neural networks ; Loss functions ; Robust regression Vědní obor RIV IN - Informatika Obor OECD Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8) CEP GA19-05704S GA ČR - Grantová agentura ČR GA18-23827S GA ČR - Grantová agentura ČR Institucionální podpora UIVT-O - RVO:67985807 DOI 10.1007/978-3-030-48791-1_43 Anotace Common types of artificial neural networks have been well known to suffer from the presence of outlying measurements (outliers) in the data. However, there are only a few available robust alternatives for training common form of neural networks. In this work, we investigate robust fitting of multilayer perceptrons, i.e. alternative approaches to the most common type of feedforward neural networks. Particularly, we consider robust neural networks based on the robust loss function of the least trimmed squares, for which we express formulas for derivatives of the loss functions. Some formulas, which are however incorrect, have been already available. Further, we consider a very recently proposed multilayer perceptron based on the loss function of the least weighted squares, which appears a promising highly robust approach. We also derive the derivatives of the loss functions, which are to the best of our knowledge a novel contribution of this paper. The derivatives may find applications in implementations of the robust neural networks, if a (gradient-based) backpropagation algorithm is used. Pracoviště Ústav informatiky Kontakt Tereza Šírová, sirova@cs.cas.cz, Tel.: 266 053 800 Rok sběru 2021 Elektronická adresa https://link.springer.com/chapter/10.1007%2F978-3-030-48791-1_43
Počet záznamů: 1