Saved in:
Bibliographic Details
Main Authors: Gao, Rui, Jaiman, Rajeev K.
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2410.04716
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914966495297536
author Gao, Rui
Jaiman, Rajeev K.
author_facet Gao, Rui
Jaiman, Rajeev K.
contents Implicit neural representations (INR) have been recently adopted in various applications ranging from computer vision tasks to physics simulations by solving partial differential equations. Among existing INR-based works, multi-layer perceptrons with sinusoidal activation functions find widespread applications and are also frequently treated as a baseline for the development of better activation functions for INR applications. Recent investigations claim that the use of sinusoidal activation functions could be sub-optimal due to their limited supported frequency set as well as their tendency to generate over-smoothed solutions. We provide a simple solution to mitigate such an issue by changing the activation function at the first layer from $\sin(x)$ to $\sin(\sinh(2x))$. We demonstrate H-SIREN in various computer vision and fluid flow problems, where it surpasses the performance of several state-of-the-art INRs.
format Preprint
id arxiv_https___arxiv_org_abs_2410_04716
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle H-SIREN: Improving implicit neural representations with hyperbolic periodic functions
Gao, Rui
Jaiman, Rajeev K.
Computer Vision and Pattern Recognition
Implicit neural representations (INR) have been recently adopted in various applications ranging from computer vision tasks to physics simulations by solving partial differential equations. Among existing INR-based works, multi-layer perceptrons with sinusoidal activation functions find widespread applications and are also frequently treated as a baseline for the development of better activation functions for INR applications. Recent investigations claim that the use of sinusoidal activation functions could be sub-optimal due to their limited supported frequency set as well as their tendency to generate over-smoothed solutions. We provide a simple solution to mitigate such an issue by changing the activation function at the first layer from $\sin(x)$ to $\sin(\sinh(2x))$. We demonstrate H-SIREN in various computer vision and fluid flow problems, where it surpasses the performance of several state-of-the-art INRs.
title H-SIREN: Improving implicit neural representations with hyperbolic periodic functions
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2410.04716