Optimization of the Local Search in the Training for SAMANN Neural Network

In this paper, we discuss the visualization of multidimensional data. A well-known procedure for mapping data from a high-dimensional space onto a lower-dimensional one is Sammon’s mapping. This algorithm preserves as well as possible all interpattern distances. We investigate an unsupervised backpropagation algorithm to train a multilayer feed-forward neural network (SAMANN) to perform the Sammon’s nonlinear projection. Sammon mapping has a disadvantage. It lacks generalization, which means that new points cannot be added to the obtained map without recalculating it. The SAMANN network offers the generalization ability of projecting new data, which is not present in the original Sammon’s projection algorithm. To save computation time without losing the mapping quality, we need to select optimal values of control parameters. In our research the emphasis is put on the optimization of the learning rate. The experiments are carried out both on artificial and real data. Two cases have been analyzed: (1) training of the SAMANN network with full data set, (2) retraining of the network when the new data points appear.     

高等数学学习方法指导的改革与实践

学习科学已经发展为一个学科群，作为学习科学的重要组成部分，高等数学学习方法指导的研究也警来越受到人们的重视，本文通过对高等数学学习方法指导的现状的剖析，提出高等数学学习方法改革的基本思路.     

Combining latent learning with dynamic programming in the modular anticipatory classifier system

Learning Classifier Systems (LCS) are rule based Reinforcement Learning (RL) systems which use a generalization capability. In this paper, we highlight the differences between two kinds of LCSs. Some are used to directly perform RL while others latently learn a model of the interactions between the agent and its environment. Such a model can be used to speed up the core RL process. Thus, these two kinds of learning processes are complementary. We show here how the notion of generalization differs depending on whether the system anticipates (like Anticipatory Classifier System (ACS) and Yet Another Classifier System (YACS)) or not (like XCS). Moreover, we show some limitations of the formalism common to ACS and YACS, and propose a new system, called Modular Anticipatory Classifier System (MACS), which allows the latent learning process to take advantage of new regularities. We describe how the model can be used to perform active exploration and how this exploration may be aggregated with the policy resulting from the reinforcement learning process. The different algorithms are validated experimentally and some limitations in presence of uncertainties are highlighted.     

Optimization schemes for decomposition of fuzzy relations

The paper proposes an optimization environment for decomposing fuzzy relations. Two basic categories of approaches are developed: (i) based on direct gradient-oriented optimization methods, and (ii) exploiting the ideas of fuzzy neurocomputation. These two approaches are carefully analyzed and contrasted making use of illustrative numerical material. Some immediate applications to pattern classification and fuzzy interpolation are studied in detail. The study includes also some generalizations of the generic decomposition problem.     

BP神经网络的一个带跳步策略的学习算法

本文利用模型逼近度的概念和跳步策略，给出了ＢＰ神经网络的一个自适应跳步学习算法，对Ｆｉｓｈｅｒ提出的分类问题和一个实际的非线性时间序列预测问题进行了计算，结果表明本文法取得了良好的效果。     

LTE网络中业务识别方案的研究与实现*1

为了实现对长期演进(Long Term Evolution,LTE)网络的业务识别,分析了S1接口用户面协议栈,利用模块化设计思想实现了对S1接口流量的业务识别.针对传统业务识别系统识别度低、统计能力不强的缺陷,在传统的业务识别系统基础上,提出了一个多识别的业务识别方案,实现了对业务类型的精确识别.经过现网数据测试验证,所设计的多识别的业务识别方案达到了预期的效果,在LTE移动通信网络业务识别领域具有推广意义.     

A frequency domain iterative learning algorithm for high-performance,periodic quadrocopter maneuvers

Quadrocopters offer an attractive platform for aerial robotic applications due to, amongst others, their hovering capability and large dynamic potential. Their high-speed flight dynamics are complex, however, and the modeling thereof has proven difficult. Control algorithms typically rely on simplified models, with feedback corrections compensating for unmodeled effects. This can lead to significant tracking errors during high-performance flight, and repeated execution typically leads to a large part of the tracking errors being repeated. This paper introduces an iterative learning scheme that non-causally compensates repeatable trajectory tracking errors during the repeated execution of periodic flight maneuvers. An underlying feedback control loop is leveraged by using its set point as a learning input, increasing repeatability and simplifying the dynamics considered in the learning algorithm. The learning is carried out in the frequency domain, and is based on a Fourier series decomposition of the input and output signals. The resulting algorithm requires little computational power and memory, and its convergence properties under process and measurement noise are shown. Furthermore, a time scaling method allows the transfer of learnt maneuvers to different execution speeds through a prediction of the disturbance change. This allows the initial learning to occur at reduced speeds, and thereby extends the applicability of the algorithm for high-performance maneuvers. The presented methods are validated in experiments, with a quadrocopter flying a figure-eight maneuver at high speed. The experimental results highlight the effectiveness of the approach, with the tracking errors after learning being similar in magnitude to the repeatability of the system.     

多核融合框架下的雷达辐射源个体识别

该文利用机器学习中多核学习(Multiple Kernel Learning,MKL)算法的多源信息融合功能,提出一种基于核层面信息融合的雷达辐射源个体识别框架。对雷达辐射源信号所提取的不同特征表示,分别构建相应的核函数或核矩阵,然后通过一定的准则计算它们的组合系数,并同时或独立获得支持向量机的分类超平面,最终实现对辐射源信号的分类。特别地,该方法能够实现辐射源信号模糊函数多个"近零"切片特征的有效融合,得到比代表性切片更优的识别性能。在3组实测雷达辐射源数据上的实验表明了所提方法的有效性。     

有机化学专业核心课程群教学质量提升改革实践

针对当前有机化学专业核心课程群教学中存在的问题,开展了相应的教学质量提升改革。通过细化课堂教学实施方案,引导学生课前主动预习;通过分段式课堂教学,提高学生课堂参与度;通过多层次的课后练习,提高学生的综合能力;通过提高考题的高阶性,引导学生注重思考能力和综合能力的发展。经过这样的教学质量提升改革,学生学习的主动性、思考能力和综合能力得到了显著的提高,为培养知识、能力和思维全面协调发展的创新人才奠定了坚实的基础。     

深度学习理论视角下的移动学习推荐系统的设计和研究

计算机技术和通信技术的迅猛发展,促使移动教学和深度学习在教育领域成为新的研究热点。首先阐明了深度学习的内涵和特征,并依此为理论基础,研究和探讨了深度学习在移动推荐系统中的应用,实现移动推荐系统的设计。