Browsing by Author "Cheng, SS"

Now showing 1 - 6 of 6

A model-selection-based self-splitting Gaussian mixture learning with application to speaker identification
(2004-12-15)
Cheng, SS
;
Wang, HM
;
Fu, HC
;
資訊工程學系
;
Department of Computer Science
We propose a self-splitting Gaussian mixture learning (SGML) algorithm for Gaussian mixture modelling. The SGML algorithm is deterministic and is able to find an appropriate number of components of the Gaussian mixture model (GMM) based on a self-splitting validity measure, Bayesian information criterion (BIC). It starts with a single component in the feature space and splits adaptively during the learning process until the most appropriate number of components is found. The SGML algorithm also performs well in learning the GMM with a given component number. In our experiments on clustering of a synthetic data set and the text-independent speaker identification task, we have observed the ability of the SGML for model-based clustering and automatically determining the model complexity of the speaker GMMs for speaker identification.
Aging influence of poly(ethylene glycol) suppressors of Cu electrolytes on gaps filling
(2006-05-01)
Liu, SH
;
Li, TC
;
Chen, C
;
Shieh, JM
;
Dai, BT
In this study, we investigate how the degradation of poly(ethylene glycol) (PEG) additives in Cu electroplating electrolytes influences the gaps filling of damascene features and the roughness of plated surfaces. The cleavage of PEG, whose reaction is enhanced by a high bias current and more plating cycles, not only diminishes the inhibition effect of the electrolytes, but also enables the formation of enormous complexes of short-chain PEG-Cu far from the reacting surfaces. Hence, the more plating cycles performed, the worse the gaps filling characteristic and the rougher the plated surfaces. Furthermore, an overshoot phenomenon on the transient cells voltage for PEG-containing electrolytes is observed and explained well by a dynamic equilibrium between PEG absorption ability and Cu reduction speed. Accordingly, the change in overshoot shape is closely related to PEG aging and this fact is employed to examine the reliability of electrolytes.
Automated information. mining on multimedia TV news archives
(2005)
Lai, PS
;
Cheng, SS
;
Sun, SY
;
Huang, TY
;
Su, JM
This paper addresses an integrated information mining techniques for multimedia TV-news archive. The utilizes techniques from the fields of acoustic, image, and video analysis, for information retrieval on news story title, newsman and scene identification. The goal is to construct a compact yet meaningful abstraction of broadcast news video, allowing users to browse through large amounts of data in a non-linear fashion with flexibility and efficiency. By using acoustic analysis, the system can classify video into news versus commercials, with 90% accuracy on a data set of 400 hours TV-news recorded off the air from July 2003 to August of 2004. By applying speaker identification and/or image detection techniques, each news stories can be segmented with an accuracy of 96%. On screen captions or subtitles are recognized by OCR techniques to produce the text title of each news stories. The extracted title words can be used to link or to navigate more related News contents on the WWW. In cooperation with facial and scene analysis and recognition techniques, OCR results can provide users with multimodality query for specific news stories. Some experimental results are presented and discussed for the system reliability and performance evaluation and comparison.
Enhanced phosphorescence and electroluminescence in triplet emitters by doping gold into cadmium selenide/zinc sulfide nanoparticles
(2005-10-01)
Liu, HW
;
Laskar, IR
;
Huang, CP
;
Cheng, JA
;
Cheng, SS
Gold-cadmium selenide/zinc sulfide (Au-CdSe/ZnS) nanocomposites (NCs) were synthesized and characterized by transmission electron microscopy (TEM), energy dispersive X-ray (EDX) analysis, ultraviolet-visible (UV-visible) absorption and photoluminescence (PL) emission spectroscopy. The PL intensity in the Au-CdSe/ZnS NCs system was found to be much greater than that of CdSe/ZnS nanoparticles (NPs) alone, because of the surface-enhanced Raman scattering of An NPs. Adding Au-CdSe/ZnS NCs to the cyclometalated iridium(Ill) complex (Ir-complex) greatly enhanced the PL intensity of a triplet emitter. Three double-layered electroluminescence (EL) devices were fabricated where the emitting zone contains the definite mixture of Ir-complex and the NCs [molar concentration of Ir-complex/NCs = 1:0 (Blank, D-1), 1: 1 (D-2) and 1:3 (D-3)] and the device D-2 exhibited optimal EL performances. (c) 2005 Elsevier B.V. All rights reserved.
Roles of additives in damascene copper electropolishing
(2006)
Liu, SH
;
Shieh, JM
;
Chen, C
;
Hensen, K
;
Cheng, SS
This study explores how the dissolution of damascene Cu depends on accelerators of organic acids in alcohol-containing H3PO4 electropolishing electrolytes. Four two-additive electrolytes that contain different accelerators, acetic, citric, citrazinic, and benzoic acids, are evaluated. At the bottom of damascene features, an esterification reaction between alcohols and organic acids efficiently forms a highly resistive layer and reduces the acidity of the solution. Accordingly, outside the damascene features, the rate of removal of Cu is dominated by the acidity of electrolytes but, inside the features, it is also determined by the resistance of the viscous layer. For a weak acidic additive, acetic acid, an extremely high additive concentration is introduced to sustain the moderate acidity of the solution to initiate intense esterification. Therefore, acetic-acid-based two-additive electrolytes exhibit excellent Cu planarization capability. (C) 2006 The Electrochemical Society.
Two-additive electrolytes for superplanarizing damascene Cu metals
(2005)
Liu, SH
;
Shieh, JM
;
Chen, C
;
Dai, BT
;
Hensen, K
Two-additive electropolishing (EP) electrolytes exhibit an extremely high planarization-efficiency in Cu damascene schemes, independent of pattern sizes (1- 50 mu m). This electrolyte is demonstrated by adding alcohols and organic acids to the H3PO4 electrolyte. The high wetting ability of alcohols allows such additives to easily access the damascene bottom. This mechanism, assisted by the reduced polishing rate associated with the high surface viscosity caused by alcohol additives, greatly passivates the damascene bottom from elecropolishing. Accordingly, the superpolishing functionality reported by Chang et al. [Electrochem. Solid-State Lett., 6, G72 (2003)] in the two-additive electrolyte outperforms in the one-additive electrolyte. (C) 2005 The Electrochemical Society.