Accelerate adversarial training with loss guided propagation for robust image classification

Adversarial training is effective to train robust image classification models. To improve the robustness, existing approaches often use many propagations to generate adversarial examples, which have high time consumption. In this work, we propose an efficient adversarial training method with loss guided propagation (ATLGP) to accelerate the adversarial training process. ATLGP takes the loss value of generated adversarial examples as guidance to control the number of propagations for each training instance at different training stages, which decreases the computation while keeping the strength of generated adversarial examples. In this way, our method can achieve comparable robustness with less time than traditional training methods. It also has good generalization ability and can be easily combined with other efficient training methods. We conduct comprehensive experiments on CIFAR10 and MNIST, the standard datasets for several benchmarks. The experimental results show that ATLGP reduces 30% to 60% training time compared with other baseline methods while achieving similar robustness against various adversarial attacks. The combination of ATLGP and ATTA (an efficient adversarial training method) achieves superior acceleration potential when robustness meets high requirements. The statistical propagation in different training stages and ablation studies prove the effectiveness of applying loss guided propagation for each training instance. The acceleration technique can more easily extend adversarial training methods to large-scale datasets and more diverse model architectures such as vision transformers.     

Bridging (Identification) divides: Ethnic minority endorsement of new diversity ideologies

In culturally diverse societies, ethnic minorities are faced with the challenge of negotiating between their national and ethnic identification. Diversity ideologies address this challenge in different ways, by prioritizing national identification in the case of assimilation, and ethnic identification in the case of multiculturalism. However, existing research has highlighted the risks and drawbacks of both ideologies, presenting polyculturalism and interculturalism as new alternatives which construe identities as more complex, dynamic, and interconnected between groups. Given that little is known about these ideologies from the minority perspective, the present study investigated their endorsement among ethnic minorities, as well as associations with their ethnic and national identification in the USA. Results show that, in general, pro-diversity ideologies (multiculturalism, interculturalism, and polyculturalism) are all supported by ethnic minorities, in contrast to assimilation. Moreover, ethnic identification is associated with support for multiculturalism, national identification is associated with support for assimilation, and both are associated with polyculturalism. For interculturalism, associations with ethnic and national identification depend on its subcomponents, which seem to address and bring together the other three ideologies. Interculturalism may therefore be a promising way forward to minimize the risks of assimilation, multiculturalism, and polyculturalism, while also maximizing their benefits to minorities and societies at large.     

基于模糊自适应变权重的经济时间序列组合预测模型研究

组合方法首先选取支持向量机预测算法和一阶指数平滑法对经济时间序列分别进行预测,来建立模糊自适应变权重组合预测模型。为对比模糊自适应变权重的经济时间序列组合预测模型的预测效果,选取了两种定值加权组合预测模型:平均加权模型、误差平方和最小组合预测模型。通过实验比较分析:模糊自适应变权重组合预测可以综合利用各单项预测方法的优点,比单一模型预测结果精度有了很大提高,且优于定值加权组合预测,在经济时间序列的预测方面有较高的应用价值。     

科学技术工作者的时间管理

科学技术工作的特点决定了时问管理是科学技术研究管理的重要方面。从时问的内涵来看,管理的效率实际上就是时间的利用效率。科学技术工作者的时间管理需遵循计划原则、节约原则和效率一效果原则;主要方法有ABC控制法、艾森豪威尔法以及时间一价值弹性分析法。时间管理还要与时俱进,这要求科学技术工作者应熟练掌握时间管理的若干技巧,深刻理解时间管理与流程再造的密切关系,还得善于利用节约时间的科学技术成果。     

基于多维空间的现代区域物流协同要素分析

运用四维空间分析方法对区域物流协同要素进行分析,指出现代区域物流协同依赖于时间、成本、效率、可持续发展等多个约束因素的协调,从而为现代区域物流管理理论与实践提供了一个系统的战略框架。     

中国餐饮业竞争力的省域差异分析

通过构建餐饮业竞争力的综合评价指标体系,利用因子分析及聚类分析,对中国各省域餐饮业竞争力进行了实证评价,分析了其差距及原因,进而提出了相应的政策建议。     

未来盈利预测信息的有用性研究

本文以2006-2008三年的年度业绩预告作为公司未来盈利预测信息的研究样本,对股权分置改革后上市公司未来盈利预测信息的有用性进行实证分析。研究表明,上市公司未来盈利预测信息具有明显的短期披露效应：公告类型与股票价格的变动方向和大小有着符号上的一致性和统计上的显著性。公告前的股价反应表明,市场存在以利好消息进行内幕交易的可能性。与以前研究结论所不同的是,长期事件窗口的结果显示盈利预测信息的披露效应存在一定的局限性。     

以更伸展的历史眼光和更宽广的世界眼光透视中国当代改革

当代中国改革是整个中国改革历史长河的一段,是世界改革历史长河的一隅。有必要建立一种＂大改革观＂,将当代中国改革放在中外改革史的经纬坐标中整体反思,走出＂两个历史宿命＂和＂两个历史循环＂。     

蜂群优化算法在带软时间窗的车辆路径问题中的应用

本文给出了带软时间窗的车辆路径问题的一种新的算法,蜂群算法.通过计算若干benchmark问题,并将结果与硬时间窗的目前最好解及蚁群算法的相应解作比较与分析,验证了算法的有效性.蜂群算法是刚刚起步的智能优化算法,目前国内外关于蜂群算法的文献较少,研究范围较窄,故本文不仅是拓宽蜂群算法应用范围的有效尝试,同时也给本身求解方法不多的软时间窗车辆路径问题提供了一种新解决方法.