Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2410.03581 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915070781423616 |
|---|---|
| author | Sun, Zicheng Zhang, Yixuan Ling, Zenan Fan, Xuhui Zhou, Feng |
| author_facet | Sun, Zicheng Zhang, Yixuan Ling, Zenan Fan, Xuhui Zhou, Feng |
| contents | Existing permanental processes often impose constraints on kernel types or stationarity, limiting the model's expressiveness. To overcome these limitations, we propose a novel approach utilizing the sparse spectral representation of nonstationary kernels. This technique relaxes the constraints on kernel types and stationarity, allowing for more flexible modeling while reducing computational complexity to the linear level. Additionally, we introduce a deep kernel variant by hierarchically stacking multiple spectral feature mappings, further enhancing the model's expressiveness to capture complex patterns in data. Experimental results on both synthetic and real-world datasets demonstrate the effectiveness of our approach, particularly in scenarios with pronounced data nonstationarity. Additionally, ablation studies are conducted to provide insights into the impact of various hyperparameters on model performance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_03581 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Nonstationary Sparse Spectral Permanental Process Sun, Zicheng Zhang, Yixuan Ling, Zenan Fan, Xuhui Zhou, Feng Machine Learning Existing permanental processes often impose constraints on kernel types or stationarity, limiting the model's expressiveness. To overcome these limitations, we propose a novel approach utilizing the sparse spectral representation of nonstationary kernels. This technique relaxes the constraints on kernel types and stationarity, allowing for more flexible modeling while reducing computational complexity to the linear level. Additionally, we introduce a deep kernel variant by hierarchically stacking multiple spectral feature mappings, further enhancing the model's expressiveness to capture complex patterns in data. Experimental results on both synthetic and real-world datasets demonstrate the effectiveness of our approach, particularly in scenarios with pronounced data nonstationarity. Additionally, ablation studies are conducted to provide insights into the impact of various hyperparameters on model performance. |
| title | Nonstationary Sparse Spectral Permanental Process |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2410.03581 |