多带抗噪声语音识别算法研究

根据Flether等人的研究,基于感知独立性假设的子带识别方法被用于抗噪声鲁棒语音识别。本文拓展子带方法,采用基于噪声污染假定的多带框架来减少噪声影响。论文不仅从理论上分析了噪声污染假定多带框架在识别性能上的潜在优势,而且提出了多带环境下的鲁棒语音识别算法。研究表明：多带框架不仅回避了独立感知假设要求,而且与子带方法相比,多带方法能更好的减少噪声影响,提高系统识别性能。     

光电对抗仿真试验系统综述

论述了研制光电对抗仿真试验系统建设的目的和意义,光电武器系统的现状和发展趋势,介绍了光电对抗仿真试验系统的一般组成,给出了光电武器系统的常规试验内容和试验流程,并论述了可能采用的试验方法。     

Synthesis of Rice Processing Plants. I. Development of Simplified Models

This article, the first of three articles on the synthesis of rice processing plants, focuses on the development of simplified mathematical models necessary for use in optimizing rice processing plants. The second concentrates on the optimal synthesis of a rice plant and the third on the sensitivity of the optimization to uncertainty in model parameters. Existing models for rice processing unit operations are not suitable for flowsheet optimization and new models need to be developed to overcome numerical difficulties that occur in optimization applications, specifically in mixed integer nonlinear programming (MINLP) applications. Simplified models of the drying, cooling, and tempering units are developed. In addition head rice yield models, used as a quality indicator, energy consumption, and economic models were also developed. Naturally, the new models exhibit some mismatch with respect to the existing models from which they were developed. However, a sensitivity analysis, presented in Part III, has shown that the optimal flowsheet structure was not sensitive to a lack of fit between the simplified and complex models. The simplified models were found adequate to be appropriate for use at the synthesis stage.     

测评软件生成系统的设计与实现

本文讨论了一个具有通用性的测试评估软件生成系统的设计思想和实现方法，给出了该生成器的系统组成、数据组织结构、查询搜索方式以及用户界面的设计。该系统具广泛的应用领域和实用意义。     

SQL的现状及其研究与发展

一、SQL简介 如果说七十年代关系数据库还处于研制实验阶段,那么到了八十年代大量关系数据库商用产品的面市几乎占据了整个数据库市场,其中的原因在于它们使用了统一的语言界面,那就是,SQL成为了关系数据库的通用语言。 1974年．IBM公司San Jose研究实验室的D．D．Chamberlin在开发SEQUEL-XRM原型时提出了SE-QUEL(Structured English Query Language),1976年又修改成为SEQUEL／2．也就是目前的SQL(StructuredQuer Language)。SEQUEL／2主要成形于IBM的Sys-tem R原型中[2],并在原有基础上大加扩展,其后逐渐完善并最终成为一种完备的结构化数据库查询语言。     

Numerical modeling of combustion processes and pollutant formations in direct-injection diesel engines

The Representative Interactive Flamelet (RIF) concept has been applied to numerically simulate the combustion processes and pollutant formation in the direct injection diesel engine. Due to the ability for interactively describing the transient behaviors of local flame structures with CFD solver, the RIF concept has the capabilities to predict the auto-ignition and subsequent flame propagation in the diesel engine combustion chamber as well as to effectively account for the detailed mechanisms of soot formation, NO_X formation including thermal NO path, prompt and nitrous NO_X formation, and reburning process. Special emphasis is given to the turbulent combustion model which properly accounts for vaporization effects on the mixture fraction fluctuations and the pdf model. The results of numerical modeling using the RIF concept are compared with experimental data and with numerical results of the commonly applied procedure which the low-temperature and high-temperature oxidation processes are represented by the Shell ignition model and the eddy dissipation model, respectively. Numerical results indicate that the RIF approach including the vaporization effect on turbulent spray combustion process successfully predicts the ignition delay time and location as well as the pollutant formation.     

Model validation and model structure determination

This paper gives a tutorial overview of basic approaches for model validation and model structure determination.Work partially supported by the Swedish Research Council for Engineering Sciences under contract 98-654.     

《宏观经济学》课程教学体系梳理

宏观经济学从资源如何充分利用这一独特的研究视角出发，以国民收入的决定为核心，以从二部门到部门到四部门为横线，从简单模型到扩展模型到总供求模型为纵线，从理论到政策为立线，构造了一个立体的理论框架体系。对宏观经济学的理论脉络进行梳理，能帮助学生提纲挈领地把握宏观经济学的理论体系。     

Extending boosting for large scale spoken language understanding

We propose three methods for extending the Boosting family of classifiers motivated by the real-life problems we have encountered. First, we propose a semisupervised learning method for exploiting the unlabeled data in Boosting. We then present a novel classification model adaptation method. The goal of adaptation is optimizing an existing model for a new target application, which is similar to the previous one but may have different classes or class distributions. Finally, we present an efficient and effective cost-sensitive classification method that extends Boosting to allow for weighted classes. We evaluated these methods for call classification in the AT&;T VoiceTone^® spoken language understanding system. Our results indicate that it is possible to obtain the same classification performance by using 30% less labeled data when the unlabeled data is utilized through semisupervised learning. Using model adaptation we can achieve the same classification accuracy using less than half of the labeled data from the new application. Finally, we present significant improvements in the “important” (i.e., higher weighted) classes without a significant loss in overall performance using the proposed cost-sensitive classification method.