Publ3 search

Z6_60GI02O0O8IDC0QEJUJ26TJDI4

Error:

Javascript is disabled in this browser. This page requires Javascript. Modify your browser's settings to allow Javascript to execute. See your browser's documentation for specific instructions.

Publikační činnost

Probíhá načítání, čekejte prosím...

Record type:	stať ve sborníku (D)
Home Department:	Ústav pro výzkum a aplikace fuzzy modelování (94410)
Title:	Fractional Loss Functions in Neural Networks: A New Perspective on Learning Dynamic
Citace	Alijani, Z. a Molek, V. Fractional Loss Functions in Neural Networks: A New Perspective on Learning Dynamic. In: 2025 IFSA World Congress NAFIPS Annual Meeting: Fuzzy Systems 60 Years Later: Past, Present, and Future 2025-08-16 Banff. Cham: Springer, 2025. ISBN 978-3-032-32831-1.

Subtitle
Publication year:	2025
Obor:
Number of pages:	12
Page from:
Page to:
Form of publication:	Elektronická verze
ISBN code:	978-3-032-32831-1
ISSN code:	2367-3370
Proceedings title:	Fuzzy Systems 60 Years Later: Past, Present, and Future
Proceedings:	Mezinárodní
Publisher name:	Springer
Place of publishing:	Cham
Country of Publication:	Sborník vydaný v zahraničí
Název konference:	2025 IFSA World Congress NAFIPS Annual Meeting
Conference venue:	Banff
Datum zahájení konference:
Typ akce podle státní příslušnosti účastníků:	Celosvětová akce
WoS code:
EID:

Key words in English:	Fractional Calculus; Loss Function; Grünwald-Letnikov Derivative
Annotation in original language:	This paper presents a new method of creating neural networkloss functions using the concepts of fractional calculus. By integratingnon-integer order derivatives, we transform conventional loss functionsinto fractional models, such as fractional Mean Squared Error, fractionalHuber loss, and fractional cross-entropy. We provide both theoreticalframeworks and empirical evidence across classification and regressionbenchmarks, emphasizing performance enhancements and practical considerations.
Annotation in english language:	This paper presents a new method of creating neural networkloss functions using the concepts of fractional calculus. By integratingnon-integer order derivatives, we transform conventional loss functionsinto fractional models, such as fractional Mean Squared Error, fractionalHuber loss, and fractional cross-entropy. We provide both theoreticalframeworks and empirical evidence across classification and regressionbenchmarks, emphasizing performance enhancements and practical considerations.

References

Reference

R01:

Complementary Content

${loading}