| Autor: |
Jeon, Sung-Ik, Kim, Seung-Gyun, Hong, SangJeen, Han, Seung-Soo |
| Zdroj: |
Advances in Neural Networks - ISNN 2010 (9783642133176); 2010, p464-471, 8p |
| Abstrakt: |
In this paper, extended Hidden Markov Model (eHMM) is employed to resolve transition detection problems in plasma etch processes using optical emission spectroscopy (OES) data. The proposed eHMM framework is a one of various semi-Markov models: a combination of semi-Markov model and segmental model. In the OES data, the endpoint is correlated to the state transition in the model. The segmental model permit adaptable modeling of data within several segments, e.g., linear, quadratic, or other regression functions. The semi-Markov model permits blending prior knowledge from previous time. The semi-Markov-Model is an extended version of the standard Hidden Markov Model (HMM), from which learning and deductive algorithms are expanded. The verification using test data is assured accuracy and excellence of the proposed eHMM in endpoint detection. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
