利用特征融合和整体多样性提升单模型鲁棒性

使用深度神经网络处理物联网设备的急剧增加产生的海量图像数据是大势所趋,但由于深度神经网络对于对抗样本的脆弱性,它容易受到攻击而危及物联网的安全.所以,如何提高模型的鲁棒性,就成了一个非常重要的课题.通常情况下,组合模型的防御表现要优于单模型防御方法,但物联网设备有限的计算能力使得组合模型难以应用.为此,提出一种在单模型上实现组合模型防御效果的模型改造及训练方法:在基础模型上添加额外的分支;使用特征金字塔对分支进行特征融合;引入整体多样性计算辅助训练.通过在MNIST和CIFAR-10这两个图像分类领域最常用的数据集上的实验表明,该方法能够显著提高模型的鲁棒性.在FGSM等4种基于梯度的攻击下的分类正确率有5倍以上的提高,在JSMA,C&W以及EAD攻击下的分类正确率可达到原模型的10倍.同时,不干扰模型对干净样本的分类精度,也可与对抗训练方法联合使用获得更好的防御效果.     

Co-evolutionary particle swarm optimization to solve constrained optimization problems

This paper presents a co-evolutionary particle swarm optimization (CPSO) algorithm to solve global nonlinear optimization problems. A new co-evolutionary PSO (CPSO) is constructed. In the algorithm, a deterministic selection strategy is proposed to ensure the diversity of population. Meanwhile, based on the theory of extrapolation, the induction of evolving direction is enhanced by adding a co-evolutionary strategy, in which the particles make full use of the information each other by using gene-adjusting and adaptive focus-varied tuning operator. Infeasible degree selection mechanism is used to handle the constraints. A new selection criterion is adopted as tournament rules to select individuals. Also, the infeasible solution is properly accepted as the feasible solution based on a defined threshold of the infeasible degree. This diversity mechanism is helpful to guide the search direction towards the feasible region. Our approach was tested on six problems commonly used in the literature. The results obtained are repeatedly closer to the true optimum solution than the other techniques.     

基于动态权值的粒子群算法的多样性分析

种群的多样性是保证粒子群优化算法收敛的前提条件,基于此提出了一个概念清晰、运算量小的多样性定义,并从粒子在寻优过程中粒子聚合程度和速度进化程度出发分析粒子群的多样性。在此基础上,提出了一种基于动态权值的改进算法,算法能自适应的调整惯性因子以保持种群多样性,有效地避免了早熟收敛。仿真实验表明该算法不仅能加快种群的进化速度,而且还能增强算法的全局收敛性,收敛概率也从15％增加到100％。     

相关瑞利衰落信道MRC接收性能分析

通过一种一般性方法研究两种天线设置时相关瑞利衰落信道上MRC接收的BPSK性能，并利用衰落信道上数字信号传输的归一化性能研究结果，分析MRC接收的MQAM的性能，给出平均比特差错率的解析表达式。     

办公建筑表皮形式的生态剖析

本文以适宜技术下生态表皮的可变性措施为切入点，结合工程实例探讨办公建筑生态表皮对建筑立面形式的多样性与动态性表达的影响，以此来解读生态策略与形式表达相结合的发展趋势。     

Performance of Broadcast and Multicast Service with Interference Cancellation Based on Macro Diversity

The cdma2000 1xEV-DO mobile communication system provides broadcast and multicast services(BCMCS) to meet an increasing demand for multimedia data services.Single frequency network(SFN) is recognized as an effective way to improve the spectral efficiency of the BCMCS system.To improve the coverage and the data rate of the BCMCS users,an interference cancellation algorithm based on macro diversity is introduced in this paper.The performance of BCMCS in CDMA2000 EV-DO networks under the condition of SFN is evaluated.The simulation results show a significant improvement in bit error ratio(BER) while quantitatively validating the efficiency of our approach.     

The effect of diversity maintenance on prediction in dynamic multi-objective optimization

There are many dynamic multi-objective optimization problems (DMOPs) in real-life engineering applications whose objectives change over time. After an environmental change occurs, prediction strategies are commonly used in dynamic multi-objective optimization algorithms to find the new Pareto optimal set (POS). Being able to make more accurate prediction means the algorithm requires fewer computational resources to make the population approximate to the Pareto optimal front (POF). This paper proposes a hybrid diversity maintenance method to improve prediction accuracy. The method consists of three steps, which are implemented after an environmental change. The first step, based on the moving direction of the center points, uses the prediction to relocate a number of solutions close to the new Pareto front. On the basis of self-defined minimum and maximum points of the POS in this paper, the second step applies the gradual search to produce some well-distributed solutions in the decision space so as to compensate for the inaccuracy of the first step, simultaneously and further enhancing the convergence and diversity of the population. In the third step, some diverse individuals are randomly generated within the region of next probable POS, which prompts the diversity of the population. Eventually the prediction becomes more accurate as the solutions with good convergence and diversity are selected after the non-dominated sort [1] on the combined solutions generated by the three steps. Compared with three other prediction methods on a series of test instances, our method is very competitive in convergence and diversity as well as the speed at which it responds to environmental changes.     

Diversity of cultivable hydrogen-producing bacteria isolated from agricultural soils, waste water sludge and cow dung

This project aimed to study the diversity of cultivable hydrogen-producing bacteria, isolated from agricultural soils, waste water sludge and cow dung by analyzing 16S rRNA gene. Isolation performed anaerobically on nutrient agar using environmental samples as inoculum yielded 106 pure isolates. These isolates were tested for their capability to produce hydrogen. Then 16S rRNA gene of the 11 isolates having such ability were PCR amplified and subjected to restriction fragment length polymorphism (RFLP) analysis in order to group them into operational taxonomic units (OTUs). RFLP was evaluated for its ability to group the 1500-1600 bp PCR products into OTUs. Isolates were presumptively identified by analysis of partial 16S rRNA gene sequence data. Each OTU was found to be phylogenetically cohesive. The 11 isolates were put into 2 main groups. Group 1 composted of WS-2-2, AS_I-1 and AS_I-2 whose partial 16S rRNA genes showed similarity of 98-99% to members of the genus Paenibacillus. Group 2 composted of WS-4-2, Lao-1-2, WS-7-11, AS-1, AS-4, CD-5 and CD-6 having highest similarity of 98-99% to members of four genera in the family Enterobacteriacea. AS_I-2 which produced the highest amount of hydrogen gas of 2.70 ml had the highest 16S rRNA gene similarity of 99% to Paenibacillus polymixa.     

日本多摩新城第15住区的实验

本文介绍日本多摩新城第15住区的开发思路、设计组织模式和设计理念。该住区的实例对中国当前的住宅开发具有参考价值。     

一种新的未知信道参数下的次最优分集合并接收机

本文推导了衰落信道中未知信道参数情况下相移键控信号多信道分集合并的最大似然序列估计算法,并在此最优方案的基础上提出了两新的次最优维持比算法,将该算法用于扩系统的ＲＡＫＥ接收机的模拟结果表明,对ＤＱＰＳＫ和编码的ＱＰＳＫ信号的接收民理想接收十分接近。