Browsing by Author "Hsu, Cheng-Wei"

Now showing 1 - 2 of 2

Thermal and angular dependence of next-generation photovoltaics under indoor lighting
(1970-01-01)
Chen, Chia-Yuan
;
Kuo, Ting-Yang
;
Huang, Chien-Wu
;
Jian, Zih-Hong
;
Hsiao, Po-Tsung
Next-generation photovoltaic technologies such as dye-sensitized solar cells, organic thin-film photovoltaics and perovskite solar cells are promising to efficiently harvest ambient light energy. However, more and deeper understanding of their photovoltaic characteristics is essential to create new applications under room light illumination. In this study, for the first time, the difference in temperature coefficients and angular dependence of photovoltaic parameters for the large-area devices are investigated systematically under the compact fluorescent lamp and light-emitting diode light. These emerging photovoltaic devices, compared with the single crystalline silicon solar cells, not only have higher open-circuit voltage (up to approximate 1 V) and better power conversion efficiency (in the range of 9.2% similar to 22.6%) but also exhibit less temperature dependent voltage and output power (< -0.6% degrees C-1), as well as broader angular response (over 75 degrees). The state-of-the-art dye-sensitized and organic thin-film devices also show advantageously positive temperature coefficients of current, and the latter even has positive thermal dependence of fill factor. These features suggest the next-generation photovoltaic devices are more favorable than the conventional crystalline silicon solar cells for real-life indoor applications.
VARIATIONAL MANIFOLD LEARNING FOR SPEAKER RECOGNITION
(2017-01-01)
Chien, Jen-Tzung
;
Hsu, Cheng-Wei
;
電機工程學系
;
Department of Electrical and Computer Engineering
This paper presents a variational manifold learning for speaker recognition based on the probabilistic linear discriminant analysis (PLDA) using i-vectors. A latent variable model is introduced to compensate the constraints of the linearity in PLDA scoring and the high dimensionality in using i-vectors. A deep variational learning is formulated to jointly optimize three objectives including a regularization for variational distributions, a reconstruction based on PLDA and a manifold learning for neighbor embedding. A stochastic gradient variational Bayesian algorithm is developed to optimize the variational lower bound of log likelihood where the expectation in the objectives is estimated via a sampling method. Interestingly, the latent variables in the proposed variational manifold PLDA (vm-PLDA) are capable of decoding or reconstructing the i-vectors. The experiments on visualization and speaker recognition show the merits of vm-PLDA in manifold learning and classification.