Aggregating multiple classification results using Choquet integral for financial distress early warning

Financial distress prediction methods based on combination classifier become a rising trend in this field. This paper applies Choquet integral to ensemble single classifiers and proposes a Choquet integral-based combination classifier for financial distress early warning. Also, as the conditions between training and pattern recognition cannot be completely consistent, so this paper proposes an adaptive fuzzy measure by using the dynamic information in the single classifier pattern recognition results which is more reasonable than the static prior fuzzy density. Finally, a comparative analysis based on Chinese listed companies’ real data is conducted to verify prediction accuracy and stability of the combination classifier. The experiment results indicate that financial distress prediction using Choquet integral-based combination classifier has higher average accuracy and stability than single classifiers.     

基于混沌加密和SVD的数字图像水印算法

为提高传统数字图像水印算法的安全性,提出一种基于混沌加密和奇异值分解的水印算法.采用Chebyshev混沌映射产生的混沌序列加密水印图像,将加密后的水印图像与原始图像进行分块,再对每一块进行奇异值分解,从而实现水印的嵌入.仿真实验结果证明,该算法对JPEG压缩、高斯噪声、椒盐噪声、滤波等具有较高的鲁棒性,同时能较好地抵抗剪切攻击.     

产学模式在《数据库原理》课程教学改革中的应用

针对目前《数据库原理》课程采用传统教学模式存在的诸多问题,就如何提高应用型本科计算机专业学生对《数据库原理》课程的学习兴趣和解决实际问题的能力,提出了产学相结合的数据库原理课程教学模式,首先让学生在工程化环境中进行相关理论学习,同时提供课程网站等辅助手段进行扩展学习,在此基础上根据企业的实际需求模拟环境让学生进行实践,并对其实践展开了进一步讨论。     

基于FPGA的AVS编码器帧内预测实现

针对帧内预测的快速算法,由于DSP架构软件顺序执行的局限性难以满足实时性要求,而FPGA以其高速的计算速度和强大的并行处理能力成为H.264和AVS编解码的理想平台。本文在FPGA平台上采用资源共享、高并行和多流水线结构实现了亮度帧内预测算法。该方法在分析AVS帧内亮度5种预测模式的基础上,将像素预测与模式判决在一个模块中完成,并且利用各模式预测的相似性,实现运算单元共享和多种模式并行执行,兼顾了处理速度和实现代价。仿真及综合结果表明该设计能够完全满足标清(704×576,30f/s)数字视频的实时处理要求。     

Numerical evaluation of the effects of friction and geometric nonlinearities on the energy release rate in three- and four-point bend end-notched flexure tests

Nonlinear finite element analyses are used to examine the effects of friction and geometric nonlinearities on the energy release rate in three- and four-point bend end-notched flexure tests. Energy release rates are first determined by a recently developed direct energy balance approach. It is shown that the finite diameter loading rollers that are typically used in practical test set-ups cause both tests to be inherently nonlinear. The effect of these nonlinearities on the energy release rate is shown to be larger in the four point than the three point test and to increase with increasing roller diameter, increasing coefficient of friction along the crack plane, and decreasing supporting span length. For the four point test, the effect of these nonlinearities is also shown to increase with increasing ratio of inner to outer span length. Next, energy release rates at the onset of crack advance are determined by a simulated compliance calibration technique. This “perceived toughness” is compared with predictions of the “true toughness” given by the direct energy balance approach at the same load. It is shown that perceived toughnesses from this simulated compliance calibration procedure are larger than previously reported results that were obtained in a similar fashion using linear theory. In addition, the perceived toughness is shown to strongly depend upon the range used for fitting the load versus deflection data to obtain compliance. These findings are used to make some general recommendations regarding use of the two test methods and their associated data reduction techniques.     

Computer supported collaborative writing in an early learning classroom

Abstract This paper describes a collaborative experiment in an early learners' classroom, equipped with special software and hardware to support the acquisition of initial reading and writing skills. This 'Computer-integrated Classroom' originated from the EU project NIMIS. Here, a new method to teach 'reading through writing' is supported by a specific software tool (T³) in the general framework of the classroom environment. Particularly, a collaborative writing task facilitated by a shared workspace system has been evaluated with a group of first graders using the T³ application. The speciality of this experiment lies in the study of domain-specific collaboration in a rich real world learning setting.     

Sliding mode control for two-time scale systems: stability issues

The paper presents some results on global exponential stability of linear time invariant systems with different time scales. The full system is decomposed in two reduced ones by means of singular perturbations and a smooth approximation of a Variable Structure Control is synthesized for each of them using Reaching Law approach. The global closed loop stability is then proved for the whole system using Lyapunov methods and a particular state space decomposition. Moreover a literal form for ε∗ (a parameter which measures the minimum separation between time scales) is derived.     

Virtual peers as partners in storytelling and literacy learning

Abstract   Literacy learning — learning how to read and write — begins long before children enter school. One of the key skills to reading and writing is the ability to represent thoughts symbolically and share them in language with an audience who may not necessarily share the same temporal and spatial context. Children learn and practice these important language skills everyday, telling stories with the peers and adults around them. In particular, storytelling in the context of peer collaboration provides a key environment for children to learn language skills important for literacy. In light of this, an embodied conversational agent, Sam , who tells stories collaboratively with children was designed. Sam looks like a peer for pre-school children, but tells stories in a developmentally advanced way, modelling narrative skills important for literacy. Results demonstrated that children who played with the virtual peer told stories that more closely resembled the virtual peer's linguistically advanced stories: using more quoted speech and temporal and spatial expressions. In addition, children listened to Sam's stories carefully, assisting her and suggesting improvements. The potential benefits of having technology play a social role in young children's literacy learning is discussed.     

Performance evaluation of continuous production lines with machines having different processing times and multiple failure modes

This paper presents an approximate analytical method for the performance evaluation of asynchronous production lines with deterministic processing times, multiple failure modes and finite buffer capacity. The discrete flow of parts is approximated by a continuous flow of material. The technique developed here is based on the approximate evaluation of a K machine line by the evaluation of K−1 two-machine lines: unlike previous work on this subject, the multiple failure approach allows to consider real and virtual failure modes for each building block in which the original line is decomposed. Indeed, to reproduce in each two-machine lines the flow of material observed in the related buffer of the original line, real failures are used to take into account the reliability of the physical machines while virtual failures are introduced to model the different interruptions of flow that can arise inside the line thus preventing machines from working. To assess the accuracy of the proposed method, extensive simulation and numerical experiments have been carried out.     

基于B/S模式的网络辅助教学系统的研究与设计

介绍了基于BS模式的网络教学系统的硬件网络拓扑结构设计以及软件的总体结构设计，并对关键子系统的设计做了说明。