EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

ReLU, Sparseness, and the Encoding of Optic Flow in Neural Networks

Oliver W Layton iD, Siyuan Peng, Scott T Steinmetz iD

Sensors · 22.11.2024 · 24 (23) · 7453

DOI: 10.3390/s24237453 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 22.11.2024. Verfasst von Oliver W Layton und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Computer Science, Colby College, Waterville, ME 04901, USA. Die Arbeit wurde bisher 11-mal zitiert (Crossref) und nennt 53 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate self-motion estimation is critical for various navigational tasks in mobile robotics. Optic flow provides a means to estimate self-motion using a camera sensor and is particularly valuable in GPS- and radio-denied environments. The present study investigates the influence of different activation functions—ReLU, leaky ReLU, GELU, and Mish—on the accuracy, robustness, and encoding properties of convolutional neural networks (CNNs) and multi-layer perceptrons (MLPs) trained to estimate self-motion from optic flow. Our results demonstrate that networks with ReLU and leaky ReLU activation functions not only achieved superior accuracy in self-motion estimation from novel optic flow patterns but also exhibited greater robustness under challenging conditions. The advantages offered by ReLU and leaky ReLU may stem from their ability to induce sparser representations than GELU and Mish do. Our work characterizes the encoding of optic flow in neural networks and highlights how the sparseness induced by ReLU may enhance robust and accurate self-motion estimation from optic flow.

Schlagwörter

Stichwörter der Autor:innen

CNNactivation functiondeep learningheadingneural networkoptic flowself-motionsparsenesssparsity

Autor:innen und Institutionen

  • Department of Computer Science, Colby College, Waterville, ME 04901, USA.
  • Microsoft Corporation, Redmond, WA 98052, USA.
  • Center for Computing Research, Sandia National Labs, Albuquerque, NM 87123, USA.

ORCID-Profile: Oliver W Layton, Scott T Steinmetz

Zitieren

APA

Layton, O. W., Peng, S., & Steinmetz, S. T. (2024). ReLU, Sparseness, and the Encoding of Optic Flow in Neural Networks. Sensors, 24(23), 7453. https://doi.org/10.3390/s24237453

Vancouver

Layton OW, Peng S, Steinmetz ST. ReLU, Sparseness, and the Encoding of Optic Flow in Neural Networks. Sensors. 2024;24(23):7453. doi:10.3390/s24237453

BibTeX
@article{Layton2024,
  author = {Layton, Oliver W and Peng, Siyuan and Steinmetz, Scott T},
  title = {ReLU, Sparseness, and the Encoding of Optic Flow in Neural Networks},
  journal = {Sensors},
  year = {2024},
  volume = {24},
  number = {23},
  pages = {7453},
  doi = {10.3390/s24237453}
}
Kostenlos starten →