Browsing by Author "Lin, CJ"

Now showing 1 - 20 of 49

A GA-based fuzzy adaptive learning control network
(2000-05-16)
Chung, IF
;
Lin, CJ
;
Lin, CT
;
電控工程研究所
;
Institute of Electrical and Control Engineering
This paper addresses the structure and an associated learning algorithm of a feedforward multilayered connectionist network for realizing the basic elements and functions of a traditional fuzzy logic controller. The proposed fuzzy adaptive learning control network (FALCON) can be contrasted with the traditional fuzzy logic control systems in their network structure and learning ability. A structure/parameter learning algorithm, called FALCON-GA, is proposed for constructing the FALCON automatically. The FALCON-GA is a three-phase hybrid learning algorithm. In the first phase, the fuzzy ART algorithm is used to do fuzzy clustering in the input/output spaces according to the supervised training data. In the second phase, the genetic algorithm (GA) is used to find proper fuzzy logic rules by associating input clusters and output clusters. Finally, in the third phase, the backpropagation algorithm is used for tuning input:output membership functions. Hence, the FALCON GA combines the backpropagation algorithm for parameter learning and both the fuzzy ART and GAs for structure learning. It can partition the input/output spaces, tune membership functions and find proper fuzzy logic rules automatically. The proposed FALCON has two important features. First, it reduces the combinatorial demands placed by the standard methods for adaptive linearization of a system. Second, the FALCON is a highly autonomous system. In its learning scheme, only the training data need to be provided from the outside world. The users need not give the initial fuzzy partitions, membership functions and fuzzy logic rules. Computer simulations have been conducted to illustrate the performance and applicability of the proposed system. (C) 2000 Elsevier Science B.V. All rights reserved.
A novel planarization of oxide-filled shallow-trench isolation
(1997-01-01)
Cheng, JY
;
Lei, TF
;
Chao, TS
;
Yen, DLW
;
Lin, CJ
Two planarization approaches of the oxide-filled trench isolation have been evaluated. Results show that the oxide-filled shallow-trench isolation technology based on a chemical-mechanical polishing (CMP) process is difficult to control and has a poor uniformity. It also results in a dishing effect in wide field regions. On the other hand, a new planarization process can achieve an excellent uniformity and fully planar surface by using a combination of a masking polysilicon layer based on a CMP process, selective wet etching for oxide refill on active regions, short-time CMP process for oxide refill, and reactive ion etching etchback. Results also show that the high breakdown yield of the gate oxide and the low leakage current of the n(+)/p junction diodes with the novel planarization process demonstrates extremely low defect density from this process. This new process is a very promising candidate for oxide-filled shallow-trench isolation.
A robust evolutionary algorithm for global optimization
(2002)
Yang, JM
;
Lin, CJ
;
Kao, CY
;
生物科技學系
;
生物資訊及系統生物研究所
This paper studies an evolutionary algorithm for global optimization. Based on family competition and adaptive rules, the proposed approach consists of global and local strategies by integrating decreasing-based mutations and self-adaptive mutations. The proposed approach is experimentally analyzed by showing that its components can integrate with one another and possess good local and global properties. Following the description of implementation details, the approach is then applied to several widely used test sets, including problems from international contests on evolutionary optimization. Numerical results indicate that the new approach performs very robustly and is competitive with other well-known evolutionary algorithms.
A speech recognition method based on the sequential multi-layer perceptrons
(1996-06-01)
Chen, WY
;
Chen, SH
;
Lin, CJ
;
交大名義發表
;
National Chiao Tung University
A novel multi-layer perceptrons (MLP)-based speech recognition method is proposed in this study. In this method, the dynamic time warping capability of hidden Markov models (HMM) is directly combined with the discriminant based learning of MLP for the sake of employing a sequence of MLPs (SMLP) as a word recognizer. Each MLP is regarded as a state recognizer to distinguish an acoustic event. Next, the word recognizer is formed by serially cascading all state recognizers. Advantages of both HMM and MLP methods are attained in this system through training the SMLP with an algorithm which combines a dynamic programming (DP) procedure with a generalized probabilistic descent (GPD) algorithm. Additionally, two sub-syllable SMLP-based schemes are studied through application of this method toward the recognition of isolated Mandarin digits. Simulation results confirm that the performance of the methods is comparable to a well modeled continuous Gaussian mixture density HMM trained with the minimum error criterion. Not only does the SMLP require less trainable parameters than the HMM system, but the former is more convenient for analysing internal features. With the aid of internal feature selection, discarding the least useful parameters of SMLP without affecting its performance is relatively easy. Copyright (C) 1996 Elsevier Science Ltd
A systolic algorithm for solving dense linear systems
(1996-12-01)
Lin, CJ
;
應用數學系
;
Department of Applied Mathematics
For an arbitrary nxn matrix A and an nx1 column vector b, we present a systolic algorithm to solve the dense linear equations Ax=b. An important consideration is that the pivot row can be changed during the execution of our systolic algorithm. The computational model consists of n linear systolic arrays. For 1 less than or equal to i less than or equal to n, the i(th) linear array:is responsible to eliminate the i(th) unknown variable x(i) of x. This algorithm requires 4n time steps to solve the linear system. The elapsed time unit within a time step is independent of the problem size n. Since the structure of a PE is simple and the same type PE executes the identical instructions, it is very suitable for VLSI implementation. The design process and correctness proof are considered in detail. Moreover,this algorithm can detect whether A is singular or not.
A systolic algorithm for solving dense linear systems
(1996-12-01)
Lin, CJ
;
交大名義發表
;
應用數學系
;
National Chiao Tung University
;
Department of Applied Mathematics
For an arbitrary nxn matrix A and an nx1 column vector b, we present a systolic algorithm to solve the dense linear equations Ax=b. An important consideration is that the pivot row can be changed during the execution of our systolic algorithm. The computational model consists of n linear systolic arrays. For 1 less than or equal to i less than or equal to n, the i(th) linear array:is responsible to eliminate the i(th) unknown variable x(i) of x. This algorithm requires 4n time steps to solve the linear system. The elapsed time unit within a time step is independent of the problem size n. Since the structure of a PE is simple and the same type PE executes the identical instructions, it is very suitable for VLSI implementation. The design process and correctness proof are considered in detail. Moreover,this algorithm can detect whether A is singular or not.
An ART-based fuzzy adaptive learning control network
(1997-11-01)
Lin, CJ
;
Lin, CT
;
交大名義發表
;
電控工程研究所
;
National Chiao Tung University
This paper addresses the structure and an associated on-line learning algorithm of a feedforward multilayered connectionist network for realizing the basic elements and functions of a traditional fuzzy logic controller. The proposed fuzzy adaptive learning control network (FALCON) can be contrasted with the traditional fuzzy logic control systems in their network structure and learning ability. An on-line structure/parameter learning algorithm called FALCON-ART is proposed for constructing the FALCON dynamically. It combines the backpropagation learning scheme for parameter learning and the fuzzy ART algorithm for structure learning. The FALCON-ART has some important features. First of all, it partitions the input state space and output control space using irregular fuzzy hyperboxes according to the distribution of training data. In many existing fuzzy or neural fuzzy control systems, the input and output spaces are always partitioned into ''grids.'' As the number of input/output variables increases, the number of partitioned grids will grow combinatorially. To avoid the problem of combinatorial growing of partitioned grids in some complex systems, the FALCON-ART partitions the input/output spaces in a flexible way based on the distribution of training data. Second, the FALCON-ART can create and train the FALCON in a highly autonomous way. In its initial form, there is no membership function, fuzzy partition, and fuzzy logic rule. They are created and begin to grow as the first training pattern arrives. Thus, the users need not give it any a priori knowledge or even any initial information on these. More notably, the FALCON-ART can on-line partition the input/output spaces, tune membership functions, find proper fuzzy logic rules, and annihilate redundant rules dynamically upon receiving online incoming training data. Computer simulations have been conducted to illustrate the performance and applicability of the proposed system.
An ART-based fuzzy adaptive learning control network
(1997-11-01)
Lin, CJ
;
Lin, CT
;
電控工程研究所
;
Institute of Electrical and Control Engineering
This paper addresses the structure and an associated on-line learning algorithm of a feedforward multilayered connectionist network for realizing the basic elements and functions of a traditional fuzzy logic controller. The proposed fuzzy adaptive learning control network (FALCON) can be contrasted with the traditional fuzzy logic control systems in their network structure and learning ability. An on-line structure/parameter learning algorithm called FALCON-ART is proposed for constructing the FALCON dynamically. It combines the backpropagation learning scheme for parameter learning and the fuzzy ART algorithm for structure learning. The FALCON-ART has some important features. First of all, it partitions the input state space and output control space using irregular fuzzy hyperboxes according to the distribution of training data. In many existing fuzzy or neural fuzzy control systems, the input and output spaces are always partitioned into ''grids.'' As the number of input/output variables increases, the number of partitioned grids will grow combinatorially. To avoid the problem of combinatorial growing of partitioned grids in some complex systems, the FALCON-ART partitions the input/output spaces in a flexible way based on the distribution of training data. Second, the FALCON-ART can create and train the FALCON in a highly autonomous way. In its initial form, there is no membership function, fuzzy partition, and fuzzy logic rule. They are created and begin to grow as the first training pattern arrives. Thus, the users need not give it any a priori knowledge or even any initial information on these. More notably, the FALCON-ART can on-line partition the input/output spaces, tune membership functions, find proper fuzzy logic rules, and annihilate redundant rules dynamically upon receiving online incoming training data. Computer simulations have been conducted to illustrate the performance and applicability of the proposed system.
Analysis of silicon nanocrystals in silicon-rich SiO2 synthesized by CO2 laser annealing - art. no. 602020
(2005)
Lin, CJ
;
Lin, GR
;
Chueh, YL
;
Chou, LJ
;
光電工程學系
The localized synthesis of 4.2-5.6 nm-Si nanocrystals (nc-Si) in Si-rich SiO2 (SRSO) by CO, laser annealing at laser intensity of below ablation-threshold (6 kW/cm(2)) is demonstrated. Since the SRSO exhibits a high absorption coefficient of up to 0.102 cm(-1) at wavelength of 10.6 mu m, a direct-writing CO, laser annealing system with focusing spot 2 size of 0.2 mm(2) is used to locally anneal the SRSO and precipitate the nc-Si. A thermophysical model reveals that the surface temperature of SRSO ranging from 130 degrees C to 3350 degrees C is achieved by varying the laser power densities from 1.5 to 13.5 kW/cm(2). The CO2 laser-ablation-threshold power density is about 6 kW/cm(2). corresponding to the optimized annealing temperature 1285 degrees C at the ablation threshold. The CO2 laser annealing is capable of the precise control on power density and spot size, which benefits from the in-situ and localized annealing temperature control of SRSO film, and also prevents from the eternal damage of the other electronic devices nearby the annealing site. The nc-Si dependent photoluminescence (PL) were observed at 806 nm or longer, whereas the laser-ablation damaged SRSO film exhibits significant blue PL at 410 nm due to the oxygen-related structural defects. The refractive index of the laser-treated SRSO film is increasing from 1.57 to 2.31 as the laser intensity increases from 1.5 to 6.0 kW/cm(2) which is mainly attributed to the increasing density of nc-Si embedded in SRSO. High resolution transmission electron microscopy (HRTEM) analysis reveals that the average size of nc-Si embedded in SRSO film is about 5.3 nm, which correlates well with the theoretical prediction of a corresponding PL at 806 nm. The HRTEM estimated square density of the nc-Si in SRSO film under the laser intensity of 6 kW/cm(2) is about 10(18) cm(-3).
Characterization and optimization of NO-nitrided gate oxide by RTP
(1995)
Sun, SC
;
Chen, CH
;
Yen, DLW
;
Lin, CJ
;
奈米中心
CO2 laser annealing synthesis of silicon nanocrystals buried in Si-rich SiO2
(2005)
Lin, CJ
;
Chueh, YL
;
Chou, LJ
;
Kuo, HC
;
Lin, GR
Localized synthesis of 3-8 nm Si nanocrystals (nc-Si) in PECVD-grown Si-rich SiO2 (SRSO) film is demonstrated using CO2 laser annealing at an intensity below the ablation-threshold (6.0 kW/cm(2)). At an optimized surface temperature of 1285 degrees C, the precipitated nc-Si in CO2-laser-annealed SRSO film results in near-infrared photoluminescence (PL) at 806 ran, whereas the ablation damage induced at higher laser intensities as well as temperatures results in blue PL at 410 nm related to structural defects. The refractive index of the laser-annealed SRSO at 633 nm increases from 1.57 to 2.31 as the laser intensity increases from 1.5 to 6.0 kW/cm(2). Transmission electron microscopy analysis reveals that the average size and volume density of Si nanocrystals embedded in the SRSO film are about 6 nm and 4.5 x 10(16) cm(-3), respectively. The CO2 laser annealing with controlled intensity and spot size can potentially accomplish in-situ, localized annealing of the SRSO film without causing irreversible damage to nearby electronics.
Correlation between surface charge accumulation and excitation intensity dependent red-shifted micro-photoluminescence of Si-implanted quartz with embedded Si nanocrystals
(2004)
Lin, CJ
;
Yu, KC
;
Kuo, HC
;
Ou-Yang, MJ
;
Lin, GR
The excitation intensity dependency of nanocrystallite Si (nc-Si) related micro-photoluminescence (mu-PL) from the multi-recipe Si-implanted quartz is characterized. The mu-PL at 724 nm contributed by nc-Si with 3-4 nm diameter is maximized after annealing at 1100degreesC for 3 hours. By increasing the excitation intensity from 10 kW/cm(2) to 300 kW/cm(2), the mu-PLs of 1-hr and 3-hr annealed quartz significantly red-shift from 723 nm to 725 nm and from 724 nm to 735 nm, respectively. This can be explained by the anomalous quantum stark effect due to a surface electric field oriented from photo-ionized nc-Si dots to quartz surface. After 1-hr illumination at power of 300 kW/cm(2), the mu-PL peak wavelength of 3-hr annealed sample is further red-shifted by 2.5 nm. By measuring the accumulated surface charges built up during optical excitation process, the correlation between excitation -intensity dependent PL wavelength red-shift and the photo-ionized nc-Si surface electric-field related quantum stark effect is primarily elucidated.
Defect-enhanced photoconductive response of silicon-implanted borosilicate glass
(2004-08-09)
Lin, GR
;
Lin, CJ
;
Lin, CK
;
光電工程學系
;
Department of Photonics
The E' (delta)-defect-enhanced photoconductivity of a metal-semiconductor-metal photodetector (MSM-PD) made on Si-implanted borosilicate glass (BSO:Si+) substrate is reported. The dark current of as-implanted BSO:Si+ MSM-PD is only 0.1 nA at bias of 70 V. The photocurrent of as-implanted BSO:Si+ MSM-PD illuminated at 488 nm is 0.91 nA, corresponding to photoconductive gain of 9.1 dB. The E' (delta)-defects luminescent at 520 nm are activated after 2 h annealing, which enhances the photocurrent of BSO:Si+ MSM-PD by one order of magnitude. Optimized responsivity, noise equivalent power, and detectivity of BSO:Si+ MSM-PD are 4.0 muA/W, 1.2x10(-9) W/Hz(1/2), and 3.5x10(5) cm Hz(1/2)/W, respectively. The electron paramagnetic resonance and etching-dependent photocurrent analysis corroborate the E' (delta)-defect-related photoconductivity of the BSO:Si+ glass. (C) 2004 American Institute of Physics.
Defect-enhanced photoconductive response of silicon-implanted borosilicate glass (vol 85, pg 935, 2004)
(2004-10-11)
Lin, GR
;
Lin, CJ
;
Lin, CK
;
光電工程學系
;
Department of Photonics
Defect-enhanced visible electroluminescence of multi-energv silicon-implanted silicon dioxide film
(2005-03-01)
Lin, CJ
;
Lin, GR
;
物理研究所
;
光電工程學系
;
Institute of Physics
White-light and blue-green electroluminescence (EL) of a multirecipe Si-ion-implanted SiO2 (SiO2:Si+) film on Si substrate are demonstrated. The blue-green photoluminescence (PL) is enhanced by the reaction of O-3 drop Si-O-Si drop O-3 --> O-3 drop Si-Si drop O-3 + O-intersticial during Si implantation. After annealing at 1100 degreesC for 180 min, the luminescence at both 415 and 455 nm is markedly enhanced by the complete activation of radiative defects, such as weak oxygen bonds, neutral oxygen vacancies (NOVs), and the precursors of nanocrystallite Si (E-delta' centers). Absorption spectroscopy and electron paramagnetic resonance confirm the existence of NOVs and E-delta' centers. The slowly rising E-delta'-related PL intensity reveals that the formation of nanocrystallite Si (nc-Si) requires longer annealing times and suggests that the activation energy for diffusion of excess Si atoms is higher than that of other defccts in ion implanted SiO2. The EL from the Ag-SiO2:Si+/n-Si-Ag metal-oxide-semiconductor diode changes from deep blue to green as the driving current increase from 0.28 to 3 A. The maximum white-light luminescent power is up to 120 nW at a bias current of 1.25 A.
Effect of electron-phonon scattering mechanisms on free-carrier absorption in quasi-one-dimensional structures
(2002-05-01)
Wu, CC
;
Lin, CJ
;
應用數學系
;
電子工程學系及電子研究所
;
Department of Applied Mathematics
The free-carrier absorption in ultrathin wires fabricated from III-V semiconductors such as n-type InSb has been investigated for the case where the electrons are scattered either by polar optical phonons or acoustic phonons. We study the interaction of longitudinal polar optical phonons with electrons and have neglected the interaction between electrons and transverse optical phonons in solids. The energy band of electrons in semiconductors is assumed to be nonparabolic. The scattering mechanisms of the interaction between electrons and phonons we consider here come from (a) electron-polar-optical-phonon scattering, (b) electron-acoustic-phonon scattering, and (c) piezoelectric scattering in semiconductors. Results are shown that the free-carrier absorption coefficient for the deformation-potential coupling is much larger than that for the piezoelectric coupling. It is also shown that the free-carrier absorption coefficient for the electron-polar-optical-phonon scattering is smaller than that for the electron-acoustic-phonon scattering. However, the free-carrier absorption coefficient increases quite slowly with the photon frequency for the electron-acoustic-phonon scattering. This is not the same result as that for the quasi-two-dimensional semiconducting structures. (C) 2002 Published by Elsevier Science B.V.
Effect of nonparabolic bands on impurity-limited mobility of semiconducting thin wires in n-type germanium
(1998)
Wu, CC
;
Lin, CJ
;
應用數學系
;
電子工程學系及電子研究所
;
Department of Applied Mathematics
The impurity-limited mobility of semiconducting thin wires for nonparabolic structures in n-type Ge has been investigated by scattering from ionized impurities with doping arsenic. The scattering is coming from the background impurities or from the uniform distribution of remote impurities. Results are shown that the impurity-limited mobility due to the scattering from background impurities increases slowly and monotonically with increasing temperature for nondegenerate semiconductors at low temperatures. This is the same qualitative property as that for an ordinary three-dimensional semiconducting structure at low temperatures. However, the mobility due to the scattering from remote impurities increases rapidly with temperature and appears much larger than that from background impurities. It is also shown that the mobility decreases rapidly and monotonically with the wire radius of semiconducting thin wire for the scattering from background impurities or remote impurities. For degenerate semiconductors, the impurity-limited mobility decreases monotonically with the wire radius for the scattering from background impurities, but the mobility does not change with the wire radius quite clearly except a discontinuous point in the neighborhood of wire radius d < 100 Angstrom for the scattering from remote impurities.
Effect of nonparabolicity on free-carrier absorption in n-type InSb films for polar optical phonon scattering
(1999-03-01)
Wu, CC
;
Lin, CJ
;
應用數學系
;
電子工程學系及電子研究所
;
Department of Applied Mathematics
The free-carrier absorption in n-type InSb films has been studied for quantum well structures fabricated from III-V semiconducting materials where the polar optical phonon scattering is predominant. We consider here two special cases: the electromagnetic radiation is polarized parallel to the layer plane and perpendicular to the layer plane separately. The energy band of electrons in semiconductors is assumed to be nonparabolic. Results show that when the electromagnetic radiation is polarized parallel to the layer plane, the free-carrier absorption coefficient is independent of temperature in a small quantum well region such as d < 30 Angstrom, but the absorption coefficient oscillates with the quantum well and depends upon the temperature in the region of larger quantum wells. When the electromagnetic radiation is polarized perpendicular to the layer plane, the dependence of the free-carrier absorption coefficient on the quantum well and temperature becomes quite complicated. (C) 1999 Elsevier Science B.V. All rights reserved.
Effect of phonon scattering on free-carrier absorption in quantum well structures
(1996-04-01)
Wu, CC
;
Lin, CJ
;
應用數學系
;
電控工程研究所
;
Department of Applied Mathematics
The free-carrier absorption has been studied for quantum well structures fabricated from III-V semiconductors such as it-type InSb films when the acoustic phonon scattering is dominant. The energy band of carriers in semiconductors is assumed to be nonparabolic. The scattering mechanisms of phonons with carriers in semiconductors are considered for the deformation-potential coupling and the piezoelectric coupling separately. Results show that the free-carrier absorption coefficient in n-type InSb films depend upon the polarization of the radiation field relative to the direction normal to the quantum well structures, the photon frequency, the film thickness and the temperature. It is also found that the free-carrier absorption coefficient could be complex due to the interaction of the radiation field and the photon field with carriers in semiconductors. Thus, the index of refraction of semiconducting films could be changed due to this carrier-phonon-photon interaction.
Effects of gamma radiation on optical fibre sensors
(2004-02-01)
Lin, KC
;
Lin, CJ
;
Lee, WY
;
光電工程學系
;
Department of Photonics
Consideration of thermo-optic coefficients is essential in the applications of optical fibre sensors and high-speed fibre communication systems. The Fabry-Perot interference method is applied to study these coefficients of four types of commercial fibres. The results are in agreement with the values calculated from the Sellmeier equation for fused silica fibre, with a difference of only about 0.2% for the Coming single-mode fibre SMF-28. In order to apply fibre sensors to a high-radiation environment, the thermo-optic coefficients for these fibres have been investigated after exposure to heavy doses of gamma radiation, and their radiation resistance has been studied.