防止过拟合的属性约简

近年来,机器学习的过拟合问题备受关注,尤其在属性约简中。为解决这一难题,提出一种融合集成策略和去除操作的算法。首先将训练模型数据平分为◢M份;然后将其中M◣-1份采用集成策略进行潜在约简计算;最后将剩余的一份进行提前测试,一旦发生过拟合则将刚加入的属性从潜在约简集中去除。利用提前测试潜在属性约简的方法来防止过拟合现象的发生,几组UCI数据的实验结果说明了新算法的有效性,同时为丰富和发展属性约简提供了一种新的方向。     

Gasification kinetics of oil sands coke

Kinetics of steam gasification of delayed and fluid cokes derived from Athabasca bitumen have been studied in 6.35 cm diameter stirred and fixed bed reactors. Experiments were carried out at atmospheric pressure and at temperatures between 800°C and 930°C. The coke particle size ranged from 0.1 to 3.5 mm, and the steam partial pressure was varied from 15.2 to 60.6 kPa. Scanning electron microscopy, surface area analysis, and mercury porosimetry were employed to relate physical structure changes in the solids to experimental kinetic data. Several kinetic models for gasification were tested with the experimental gasification data. Rate expressions are presented for carbon conversion and for the release of sulphur.     

深圳市大康河综合整治工程生态化治理研究

以大康河综合整治工程为背景对河道生态化整治进行研究,对工程涉及的防洪、水质、景观等基本现状进行评价,提出了生态治理的指导思想,并具体介绍了防洪、水质、景观等综合治理措施,通过对大康河的综合整治,使得工程既能满足防洪要求又能实现良好的生态景观效果。     

Random Forests with ensemble of feature spaces

Random Forests receive much attention from researchers because of their excellent performance. As Breiman suggested, the performance of Random Forests depends on the strength of the weak learners in the forests and the diversity among them. However, in the literature, many researchers only considered pre-processing of the data or post-processing of the Random Forests models. In this paper, we propose a new method to increase the diversity of each tree in the forests and thereby improve the overall accuracy. During the training process of each individual tree in the forest, different rotation spaces are concatenated into a higher space at the root node. Then the best split is exhaustively searched within this higher space. The location where the best split lies decides which rotation method to be used for all subsequent nodes. The performance of the proposed method here is evaluated on 42 benchmark data sets from various research fields and compared with the standard Random Forests. The results show that the proposed method improves the performance of the Random Forests in most cases.     

Large-scale SRAM-based IP lookup architectures using compact trie search structures

SRAM-based pipelined architectures for high-speed IP lookup using Field Programmable Gate Arrays (FPGAs) has recently attracted a great deal of attention from researchers. Due to the limited amount of on-chip memory and the number of I/O pins of FPGAs, compact data structures providing high memory efficiency are in great demand.     

Discovering small-world in association link networks for association learning

Association Link Network (ALN) is a kind of Semantic Link Network built by mining the association relations among multimedia Web resources for effectively supporting Web intelligent application such as Web-based learning, and semantic search. This paper explores the Small-World properties of ALN to provide theoretical support for association learning (i.e., a simple idea of “learning from Web resources”). First, a filtering algorithm of ALN is proposed to generate the filtered status of ALN, aiming to observe the Small-World properties of ALN at given network size and filtering parameter. Comparison of the Small-World properties between ALN and random graph shows that ALN reveals prominent Small-World characteristic. Then, we investigate the evolution of Small-World properties over time at several incremental network sizes. The average path length of ALN scales with the network size, while clustering coefficient of ALN is independent of the network size. And we find that ALN has smaller average path length and higher clustering coefficient than WWW at the same network size and network average degree. After that, based on the Small-World characteristic of ALN, we present an Association Learning Model (ALM), which can efficiently provide association learning of Web resources in breadth or depth for learners.     

FAULT DIAGNOSIS OF HYDRAULIC SERVO SYSTEM USING THE UNSCENTED KALMAN FILTER

Fault occurrence can be embodied by the physical parameter variations of the hydraulic servo system. Faults can, therefore, be diagnosed according to the model coefficient variations of the hydraulic servo system. This paper proposes an approach for fault diagnosis based on the unscented Kalman filter (UKF) with a mathematical model of the hydraulic servo system. The mathematical model is established using the dynamic equations of the hydraulic servo system. Based on the fault mechanism analysis results, several important system model parameters that can separately represent different faults in different components of the hydraulic servo system are chosen. Discrete state space equations are derived from the dynamic equations. The UKF algorithm is used to estimate the important system model parameters of the hydraulic servo system by utilizing the discretized state space model. According to the variations of these model parameters, the fault modes and locations of the hydraulic servo system can be diagnosed and isolated. Two types of faults, namely, abrupt fault in servovalve gain and slow wear fault in hydraulic cylinder piston, which cannot be directly detected from the system output, are introduced individually to the hydraulic servo system in this work. By comparing with the extended Kalman Filter, three different experimental cases are used to validate the effectiveness of the UKF for hydraulic servo system fault diagnosis.     

Authentification and identification — Taking the user into account

Username/Password is still the prevailing authentication mechanism for internet based services — but it is not secure! We show how new authentication and identification mechanisms focused on usability and security can change this.     

SSI:一种IPv6/IPv4多址同源识别模型

IPv6/IPv4共存环境下多址同源识别是共存网络管理与拓扑发现的一个关键问题。现有研究主要集中于子网内部的双栈发现及单一IP协议栈中的别名解析,难以识别远程IPv6/IPv4共存网络中的多址同源。通过分析同源地址间的本质联系,提出一种IPv6/IPv4多址同源识别模型(SSI),该模型综合利用特殊地址格式匹配、TCP时钟指纹比对和上层协议短时致瘫等多种方式来提高同源地址的识别能力。实验结果表明,上述方法均可有效识别IPv6/IPv4多址同源;SSI模型具有较理想的识别率和正确率。     

Development of high-performance liquid chromatography criteria for determination of grades of commercial olive oils. Part I. The normal ranges for the triacylglycerols

Criteria for authentic olive oils were developed from isocratic high-performance liquid chromatography analyses of 99 olive oils from the major Mediterranean producers in the 1983–1986 crop years. Authentic olive oils include extra virgin, virgin and pure or refined oils, but exclude all reesterified and adulterated oils. The extra virgin through pure grades will have a combined area for the LOO (C_18:2C_18:1C_18:1), LOP (C_18:2C_18:1C_16:0), OOO (C_18:1C_18:1C_18:1), POO (C_16:0C_18:1C_18:1), POP (C_16:0C_18:1C_18:1), and SOO (C_18:0C_18:1C_18:1) peaks between 82.0 and 92.6% of the total area (L, linoleic; O, oleic; P, palmitic; S, stearic). Authentic oils will have ratios of LOO/LOP and OOO/POO that coincide with a line defined by OOO/POO=0.7844(LOO/LOP)+0.0968; correlation coefficient is 0.885. Authentic oils will not have a trilinolein (LLL) peak over 0.5% in area. Neither triolein (OOO) nor any other single peak suffices to characterize an olive oil sample as one of the authentic grades.