|
Regularized Kernel Forms of Minimum Squared Error Method
XU Jian-hua, ZHANG Xue-gong, LI Yan-da
Front. Electr. Electron. Eng.. 2006, 1 (1): 1-7.
https://doi.org/10.1007/s11460-005-0011-y
Minimum squared error (MSE) algorithm is one of the classical pattern recognition and regression analysis methods, whose objective is to minimize the squared error summation between the output of linear function and the desired output. In this paper, the MSE algorithm is modified by using kernel functions satisfying the Mercer condition and regularization technique; and the nonlinear MSE algorithms based on kernels and regularization term, that is, the regularized kernel forms of MSE algorithm, are proposed. Their objective functions include the squared error summation between the output of nonlinear function based on kernels and the desired output and a proper regularization term. The regularization technique can handle ill-posed problems, reduce the solution space, and control the generalization. Three squared regularization terms are utilized in this paper. In accordance with the probabilistic interpretation of regularization terms, the difference among three regularization terms is given in detail. The synthetic and real data are used to analyze the algorithm performance.
Related Articles |
Metrics
|
|
Open-Set Face Verification Algorithm Using Competitive Negative Samples
YANG Qiong, DING Xiao-qing
Front. Electr. Electron. Eng.. 2006, 1 (1): 20-25.
https://doi.org/10.1007/s11460-005-0012-x
A novel face verification algorithm using competitive negative samples is proposed. In the algorithm, the tested face matches not only with the claimed client face but also with competitive negative samples, and all the matching scores are combined to make a final decision. Based on the algorithm, three schemes, including closestnegative-sample scheme, all-negative-sample scheme, and closest-few-negative-sample scheme, are designed. They are tested and compared with the traditional similaritybased verification approach on several databases with different features and classifiers. Experiments demonstrate that the three schemes reduce the verification error rate by 25.15%, 30.24%, and 30.97%, on average, respectively.
Related Articles |
Metrics
|
|
Novel Quantum Genetic Algorithm and Its Applications
ZHANG Ge-xiang, LI Na, JIN Wei-dong, HU Lai-zhao
Front. Electr. Electron. Eng.. 2006, 1 (1): 31-36.
https://doi.org/10.1007/s11460-005-0014-8
By introducing strong parallelism of quantum computing into evolutionary algorithm, a novel quantum genetic algorithm (NQGA) is proposed. In NQGA, a novel approach for updating the rotation angles of quantum logic gates and a strategy for enhancing search capability and avoiding premature convergence are adopted. Several typical complex continuous functions are chosen to test the performance of NQGA. Also, NQGA is applied in selecting the best feature subset from a large number of features in radar emitter signal recognition. The testing and experimental results of feature selection show that NQGA presents good search capability, rapid convergence, short computing time, and ability to avoid premature convergence effectively.
Related Articles |
Metrics
|
|
Generalized Maximum Likelihood Algorithm for Direction-of-Arrival Estimation of Coherent Sources
WANG Bu-hong, WANG Yong-liang, CHEN Hui, GUO Ying
Front. Electr. Electron. Eng.. 2006, 1 (1): 42-47.
https://doi.org/10.1007/s11460-005-0009-5
The generalized maximum likelihood (GML) algorithm for direction-of-arrival estimation is proposed. Firstly, a new data model is established based on generalized steering vectors and generalized array manifold matrix. The GML algorithm is then formulated in detail. It is flexible in the sense that the arriving sources may be a mixture of multiclusters of coherent sources, the array geometry is unrestricted, and the number of sources resolved can be larger than the number of sensors. Secondly, the comparison between the GML algorithm and the conventional deterministic maximum likelihood (DML) algorithm is presented based on their respective geometrical interpretation. Subsequently, the estimation consistency of GML is proved, and the estimation variance of GML is derived. It is concluded that the performance of the GML algorithm coincides with that of the DML algorithm in the incoherent sources case, while it improves greatly in the coherent source case. By using genetic algorithm, GML is realized, and the simulation results illustrate its improved performance compared with DML, especially in the case of multiclusters of coherent sources.
Related Articles |
Metrics
|
|
Analysis and Design of Hybrid Excitation Permanent Magnet Synchronous Generators
JIN Wan-bing, ZHANG Dong, AN Zhong-liang, TAN Ren-yuan
Front. Electr. Electron. Eng.. 2006, 1 (1): 63-66.
https://doi.org/10.1007/s11460-005-0006-8
On the basis of a conventional permanent magnet (PM) synchronous generator s construction, a novel kind of Hybrid Excitation Permanent Magnet Synchronous Generator (HEPMSG) is introduced by inserting exciting winding in the stator or rotor. Firstly, the construction of HEPMSG is improved with the addition of PM excitation on the ferromagnetic pole, and its working principle and design method are studied in detail. Then, an appropriate exciting current control system is presented considering the characteristics of HEPMSG. Finally, a prototype is made, and test results confirm the analysis and design.
Related Articles |
Metrics
|
|
Accelerometer Design Using MOS Ring Oscillator
ZHANG Zhao-hua, YUE Rui-feng, LIU Li-tian
Front. Electr. Electron. Eng.. 2006, 1 (1): 77-81.
https://doi.org/10.1007/s11460-005-0015-7
A digital accelerometer is developed by using a ring oscillator (RO) and a mixer. The sensitive unit of the accelerometer is the metal-oxide semiconductor (MOS) ROs located on silicon beams. Based on the piezoresistive effect of the MOS RO, the accelerometer transduces the acceleration into frequency output. The syntonic frequency of the MOS RO changes in relation to many environmental elements, such as temperature, source voltage, and so on. The mixer is an interior signal processor that improves the output signal characteristics, with the digital output signal representing the frequency change. As the accelerometer is based on the piezoresistive effect of the MOS RO, the frequency characteristics of the MOS RO and its relationship with the acceleration are described. The device has been fabricated using standard integrated circuits processing methods combined with the Micro-Electro-Mechanical Systems process. The characteristics of the sample chip are in agreement with the theoretical predictions. The accelerometer has a high sensitivity of 6.91 kHz/g, a low temperature coefficient, and a simple fabrication process.
Related Articles |
Metrics
|
|
Intermittency in Switching Power Converters: Theoretical Analysis
ZHOU Yu-fei, CHEN Jun-ning, KE Dao-ming, TSE Chi K., QIU Shui-sheng, SHI Long-xing, SUN Wei-feng
Front. Electr. Electron. Eng.. 2006, 1 (1): 92-97.
https://doi.org/10.1007/s11460-005-0024-6
In view of reasonable explanation of intermittent subharmonics and chaos that can be gained from coupling filter between circuits, this paper discusses a method that maps time bifurcation with parameter bifurcation. Based on this mapping method, the general analysis method of characteristic multiplier, which is originally aimed at parameter bifurcation, can be used for the study of intermittency, i.e., time bifurcation. In this paper, all researches coming from characteristic multipliers, parameter-bifurcation diagrams, and the largest Lyapunov exponent indicate the same results as those produced by simulation and experiment. Thus, it is proved theoretically that the intermittency in switching power converter can be explained in terms of coupling of spurious interference.
Related Articles |
Metrics
|
23 articles
|