Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2507.10632 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913941013135360 |
|---|---|
| author | Saito, Issei Nagano, Masatoshi Nakamura, Tomoaki Mochihashi, Daichi Mimura, Koki |
| author_facet | Saito, Issei Nagano, Masatoshi Nakamura, Tomoaki Mochihashi, Daichi Mimura, Koki |
| contents | In this paper, we propose RFF-GP-HSMM, a fast unsupervised time-series segmentation method that incorporates random Fourier features (RFF) to address the high computational cost of the Gaussian process hidden semi-Markov model (GP-HSMM). GP-HSMM models time-series data using Gaussian processes, requiring inversion of an N times N kernel matrix during training, where N is the number of data points. As the scale of the data increases, matrix inversion incurs a significant computational cost. To address this, the proposed method approximates the Gaussian process with linear regression using RFF, preserving expressive power while eliminating the need for inversion of the kernel matrix. Experiments on the Carnegie Mellon University (CMU) motion-capture dataset demonstrate that the proposed method achieves segmentation performance comparable to that of conventional methods, with approximately 278 times faster segmentation on time-series data comprising 39,200 frames. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_10632 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Scalable Unsupervised Segmentation via Random Fourier Feature-based Gaussian Process Saito, Issei Nagano, Masatoshi Nakamura, Tomoaki Mochihashi, Daichi Mimura, Koki Machine Learning Artificial Intelligence In this paper, we propose RFF-GP-HSMM, a fast unsupervised time-series segmentation method that incorporates random Fourier features (RFF) to address the high computational cost of the Gaussian process hidden semi-Markov model (GP-HSMM). GP-HSMM models time-series data using Gaussian processes, requiring inversion of an N times N kernel matrix during training, where N is the number of data points. As the scale of the data increases, matrix inversion incurs a significant computational cost. To address this, the proposed method approximates the Gaussian process with linear regression using RFF, preserving expressive power while eliminating the need for inversion of the kernel matrix. Experiments on the Carnegie Mellon University (CMU) motion-capture dataset demonstrate that the proposed method achieves segmentation performance comparable to that of conventional methods, with approximately 278 times faster segmentation on time-series data comprising 39,200 frames. |
| title | Scalable Unsupervised Segmentation via Random Fourier Feature-based Gaussian Process |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2507.10632 |