Unveiling the Potential of Superexpressive Networks in Implicit Neural Representations
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915215225913344 |
|---|---|
| author | Mudiyanselage, Uvini Balasuriya Cho, Woojin Jo, Minju Park, Noseong Lee, Kookjin |
| author_facet | Mudiyanselage, Uvini Balasuriya Cho, Woojin Jo, Minju Park, Noseong Lee, Kookjin |
| contents | In this study, we examine the potential of one of the ``superexpressive'' networks in the context of learning neural functions for representing complex signals and performing machine learning downstream tasks. Our focus is on evaluating their performance on computer vision and scientific machine learning tasks including signal representation/inverse problems and solutions of partial differential equations. Through an empirical investigation in various benchmark tasks, we demonstrate that superexpressive networks, as proposed by [Zhang et al. NeurIPS, 2022], which employ a specialized network structure characterized by having an additional dimension, namely width, depth, and ``height'', can surpass recent implicit neural representations that use highly-specialized nonlinear activation functions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_21166 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Unveiling the Potential of Superexpressive Networks in Implicit Neural Representations Mudiyanselage, Uvini Balasuriya Cho, Woojin Jo, Minju Park, Noseong Lee, Kookjin Machine Learning In this study, we examine the potential of one of the ``superexpressive'' networks in the context of learning neural functions for representing complex signals and performing machine learning downstream tasks. Our focus is on evaluating their performance on computer vision and scientific machine learning tasks including signal representation/inverse problems and solutions of partial differential equations. Through an empirical investigation in various benchmark tasks, we demonstrate that superexpressive networks, as proposed by [Zhang et al. NeurIPS, 2022], which employ a specialized network structure characterized by having an additional dimension, namely width, depth, and ``height'', can surpass recent implicit neural representations that use highly-specialized nonlinear activation functions. |
| title | Unveiling the Potential of Superexpressive Networks in Implicit Neural Representations |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2503.21166 |