Abstract
The first step to realize an automatic experimental data analysis for fusion plasma experiments is fitting noisy data of temperature and density spatial profiles, which are routinely obtained. However, it has been difficult to construct algorithms that fit all the data without overfitting and underfitting. In this paper, we show that this difficulty originates from the lack of knowledge of the probability distribution that the measurement data follow. We demonstrate the use of a machine learning technique to estimate the data distribution and to construct an optimal generative model. We show that the fitting algorithm based on the generative modeling outperforms classical heuristic methods in terms of the stability as well as the accuracy.
| Original language | English |
|---|---|
| Article number | 8745539 |
| Pages (from-to) | 3305-3314 |
| Number of pages | 10 |
| Journal | IEEE Transactions on Plasma Science |
| Volume | 47 |
| Issue number | 7 |
| DOIs | |
| State | Published - Jul 2019 |
| Externally published | Yes |
Funding
Manuscript received August 22, 2018; revised January 29, 2019; accepted April 10, 2019. Date of current version July 9, 2019. This work was supported in part by the Yazaki Memorial Foundation for Science and Technology and in part by the Grant of Joint Research by the National Institutes of Natural Sciences (NINS). The review of this paper was arranged by Senior Editor E. Surrey. (Corresponding author: K. Fujii.) K. Fujii and M. Hasuo are with the Department of Mechanical Engineering and Science, Graduate School of Engineering, Kyoto University, Kyoto 615-8540, Japan (e-mail: [email protected]).
Keywords
- Automatic analysis
- Bayesian method
- neural network
- variational inference