Conditional expectation of integrands and random sets

The conditional expectation of integrands and random sets is the main tool of stochastic optimization. This work wishes to make up for the lack of real synthesis about this subject. We improve the existing hypothesis and simplify the corresponding proofs. In the convex case we especially study the problem of the exchange of conditional expectation and subdifferential operators.     

Fluorescence line narrowed spectra of Zn and metal-free cytochrome c

Fluorescence line narrowing (FLN) spectroscopy was used to study the role of the polypeptide chain in influencing the spectrum of Zn-substituted cytochrome c (Zn cyt c) and metal-free cyt c (porphyrin cyt c). For both derivatives the spectra show characteristics of relaxed fluorescence from an inhomogeneously broadened sample. Zero phonon lines and phonon wings can be clearly distinguished, and vibrational frequencies of the ground and excited states were identified. The inhomogeneous distribution width for porphyrin cyt c is slightly wider than that of Zn cyt c and a second population of molecules was apparent in the porphyrin cyt c. The phonon coupling was greater for Zn cyt c than for porphyrin cyt c, which may be due to the extra coupling to the polypeptide chain by metal ligation.This work has been done in partial fulfillment of the requirements for the Doctor of Philosophy degree.     

二维随机激光器的模式选择及阈值与饱和特性

基于将Maxwell方程与四能级原子系统速率方程相结合而建立起随机激光时域理论，并利用有限时域差分法，研究了二维随机介质中激光模式的输出特性与介质尺寸、外形及抽运速率等参数的关系.结果表明，与传统激光模式相似，随机激光模式的强度随抽运速率的变化不仅具有阈值特性，而且具有饱和特性.基于模式特性对介质及抽运参数的依赖关系，提出了二维随机激光器的选模方式，在很大程度上不同于传统激光器的选模方式. 关键词： 随机激光器 模式选择 无序介质中的光学特性     

Feature Selection using Fuzzy Support Vector Machines

The feature selection consists of obtaining a subset of these features to optimally realize the task without the irrelevant ones. Since it can provide faster and cost-effective learning machines and also improve the prediction performance of the predictors, it is a crucial step in machine learning. The feature selection methods using support machines have obtained satisfactory results, but the noises and outliers often reduce the performance. In this paper, we propose a feature selection approach using fuzzy support vector machines and compare it with the previous work, the results of experiments on the UCI data sets show that feature selection using fuzzy SVM obtains better results than using SVM.     

Electrochemical Immunoanalysis for Carcinoembryonic Antigen Based on Multilayer Architectures of Gold Nanoparticles and Polycation Biomimetic Interface on Glassy Carbon Electrode

This paper describes a layer‐by‐layer (LBL) self‐assembly process of chitosan (CTS) and gold nanoparticles (Au) on the pretreated negatively charged glassy carbon (GC) electrode to fabricate electrochemistry immunosensor with a nontoxic biomimetic interface, which provided an environment similar to a native system and allowed more freedom in orientation for immobilization of carcinoembryonic antibody (anti‐CEA) to monitor carcinoembryonic antigen (CEA). UV‐vis spectroscope, atomic force microscopy (AFM), and cyclic voltammetric (CV) measurements were used to follow the multilayer film formation. The performance of the biominetic interface and factors influencing the assay system were investigated in detail. The differential pulse voltammetry (DPV) current response is used for the CEA concentration assay. The dynamic range was from 0.50 to 80.00 ng mL^?1 with a detection limit of 0.27 ng mL^?1 at 3σ. In addition, the experiment results indicate that immobilization described in this proposed method exhibits a good sensitivity, selectivity, and stability.     

Automatic detection and identification of shocks in Gaussian state‐space models: a Bayesian approach

An automatic monitoring and intervention algorithm that permits the supervision of very general aspects in an univariate linear Gaussian state–space model is proposed. The algorithm makes use of a model comparison and selection approach within a Bayesian framework. In addition, this algorithm incorporates the possibility of eliminating earlier interventions when subsequent evidence against them comes to light. Finally, the procedure is illustrated with two empirical examples taken from the literature. Copyright © 2005 John Wiley & Sons, Ltd.     

Bayesian Object Identification: Variants

We present a Bayesian theory of object identification. Here, identifying an object means selecting a particular observation from a group of observations (variants), this observation (the regular variant) being characterized by a distributional model. In this sense, object identification means assigning a given model to one of several observations. Often, it is the statistical model of the regular variant, only, that is known. We study an estimator which relies essentially on this model and not on the characteristics of the “irregular” variants. In particular, we investigate under what conditions this variant selector is optimal. It turns out that there is a close relationship with exchangeability and Markovian reversibility. We finally apply our theory to the case of irregular variants generated from the regular variant by a Gaussian linear model.     

A new algorithm for fixed design regression and denoising

In this paper, we present a new algorithm to estimate a regression function in a fixed design regression model, by piecewise (standard and trigonometric) polynomials computed with an automatic choice of the knots of the subdivision and of the degrees of the polynomials on each sub-interval. First we give the theoretical background underlying the method: the theoretical performances of our penalized least-squares estimator are based on non-asymptotic evaluations of a mean-square type risk. Then we explain how the algorithm is built and possibly accelerated (to face the case when the number of observations is great), how the penalty term is chosen and why it contains some constants requiring an empirical calibration. Lastly, a comparison with some well-known or recent wavelet methods is made: this brings out that our algorithm behaves in a very competitive way in term of denoising and of compression.