Neural Tangent Kernel of Neural Networks with Loss Informed by Differential Operators
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_ | 1866913735993458688 |
|---|---|
| author | Gan, Weiye Li, Yicheng Lin, Qian Shi, Zuoqiang |
| author_facet | Gan, Weiye Li, Yicheng Lin, Qian Shi, Zuoqiang |
| contents | Spectral bias is a significant phenomenon in neural network training and can be explained by neural tangent kernel (NTK) theory. In this work, we develop the NTK theory for deep neural networks with physics-informed loss, providing insights into the convergence of NTK during initialization and training, and revealing its explicit structure. We find that, in most cases, the differential operators in the loss function do not induce a faster eigenvalue decay rate and stronger spectral bias. Some experimental results are also presented to verify the theory. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_11029 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Neural Tangent Kernel of Neural Networks with Loss Informed by Differential Operators Gan, Weiye Li, Yicheng Lin, Qian Shi, Zuoqiang Machine Learning Spectral bias is a significant phenomenon in neural network training and can be explained by neural tangent kernel (NTK) theory. In this work, we develop the NTK theory for deep neural networks with physics-informed loss, providing insights into the convergence of NTK during initialization and training, and revealing its explicit structure. We find that, in most cases, the differential operators in the loss function do not induce a faster eigenvalue decay rate and stronger spectral bias. Some experimental results are also presented to verify the theory. |
| title | Neural Tangent Kernel of Neural Networks with Loss Informed by Differential Operators |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2503.11029 |