EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-SA 4.0

On the Use of the Novel Zorro Activation Functions in Long Short-Term Memory Neural Networks

Victor Adrian Jimenez, Matías Roodschild, Jorge Gotay-Sardiñas, Adrián Will

Journal of Computer Science and Technology · 10.04.2026 · 26 (1) · e03

DOI: 10.24215/16666038.26.e03 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Computer Science and Technology, online erschienen am 10.04.2026. Verfasst von Victor Adrian Jimenez und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-SA 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Activation functions are fundamental components of modern neural networks, including Large Language Models (LLMs). Nonlinear activations regulate the flow of information in most models and determine how data is processed. However, training or even fine-tuning very large and complex models with such activations, like those used in ChatGPT and DeepSeek, remains out of reach for many researchers. For this reason, traditional architectures like LSTM remain relevant for moderate or small-scale applications. In this context, the choice of appropriate functions can significantly influence a network's ability to learn complex patterns under limited computing resources. In this work, we analyze the behavior of a novel family of activation functions, called Zorro, which can serve as gating functions in LSTM architectures. We propose replacing traditional activation and gating functions in LSTMs with Zorro functions to improve model performance and convergence speed. Unlike conventional approaches, our method assigns a different function to each gate or activation, enabling the application of the same methodology to any gated architecture. We evaluate the modified LSTM models on widely used small-scale benchmark datasets, including Japanese Vowels and Human Activity Recognition for classification, as well as Chickenpox and Turbofan Degradation for regression. The results show that our method improves model accuracy by up to 10% and reduces training time by up to 15%.

Zitieren

APA

Jimenez, V. A., Roodschild, M., Gotay-Sardiñas, J., & Will, A. (2026). On the Use of the Novel Zorro Activation Functions in Long Short-Term Memory Neural Networks. Journal of Computer Science and Technology, 26(1), e03. https://doi.org/10.24215/16666038.26.e03

Vancouver

Jimenez VA, Roodschild M, Gotay-Sardiñas J, Will A. On the Use of the Novel Zorro Activation Functions in Long Short-Term Memory Neural Networks. Journal of Computer Science and Technology. 2026;26(1):e03. doi:10.24215/16666038.26.e03

BibTeX
@article{Jimenez2026,
  author = {Jimenez, Victor Adrian and Roodschild, Matías and Gotay-Sardiñas, Jorge and Will, Adrián},
  title = {On the Use of the Novel Zorro Activation Functions in Long Short-Term Memory Neural Networks},
  journal = {Journal of Computer Science and Technology},
  year = {2026},
  volume = {26},
  number = {1},
  pages = {e03},
  doi = {10.24215/16666038.26.e03}
}
Kostenlos starten →