Reliable Space Pursuing for Reliability-based Design Optimization with Black-box Performance Functions

Reliability-based design optimization （RBDO） is intrinsically a double-loop procedure since it involves an overall optimization and an iterative reliability assessment at each search point. Due to the double-loop procedure, the computational expense of RBDO is normally very high. Current RBDO research focuses on problems with explicitly expressed performance functions and readily available gradients. This paper addresses a more challenging type of RBDO problem in which the performance functions are computation intensive. These computation intensive functions are often considered as a ＂black-box＂ and their gradients are not available or not reliable. On the basis of the reliable design space （RDS） concept proposed earlier by the authors, this paper proposes a Reliable Space Pursuing （RSP） approach, in which RDS is first identified and then gradually refined while optimization is performed. It fundamentally avoids the nested optimization and probabilistic assessment loop. Three well known RBDO problems from the literature are used for testing and demonstrating the effectiveness of the proposed RSP method.

AN ALGORITHM FOR AUTOMATICALLY GENERATING BLACK-BOX TEST CASES

Selection of test cases plays a key role in improving testing efficiency. Black-box testing is an important way of testing, and its validity lies on the selection of test cases in some sense. A reasonable and effective method about the selection and generation of test cases is urgently needed. This letter first introduces some usualmethods on black-box test case generation,then proposes a new algorithm based on interface parameters and discusses its properties, finally shows the effectiveness of the algorithm.

A Fast Two-Stage Black-Box Deep Learning Network Attacking Method Based on Cross-Correlation

Deep learning networks are widely used in various systems that require classification.However,deep learning networks are vulnerable to adversarial attacks.The study on adversarial attacks plays an important role in defense.Black-box attacks require less knowledge about target models than white-box attacks do,which means black-box attacks are easier to launch and more valuable.However,the state-of-arts black-box attacks still suffer in low success rates and large visual distances between generative adversarial images and original images.This paper proposes a kind of fast black-box attack based on the cross-correlation(FBACC)method.The attack is carried out in two stages.In the first stage,an adversarial image,which would be missclassified as the target label,is generated by using gradient descending learning.By far the image may look a lot different than the original one.Then,in the second stage,visual quality keeps getting improved on the condition that the label keeps being missclassified.By using the cross-correlation method,the error of the smooth region is ignored,and the number of iterations is reduced.Compared with the proposed black-box adversarial attack methods,FBACC achieves a better fooling rate and fewer iterations.When attacking LeNet5 and AlexNet respectively,the fooling rates are 100%and 89.56%.When attacking them at the same time,the fooling rate is 69.78%.FBACC method also provides a new adversarial attack method for the study of defense against adversarial attacks.

<i>PP</i>and <i>P<span style='text-decoration:overline;'>P</span></i>Multi-Particles Production Investigation Based on CCNN Black-Box Approach

The multiplicity distribution (P(nch)) of charged particles produced in a high energy collision is a key quantity to understand the mechanism of multiparticle production. This paper describes the novel application of an artificial neural network (ANN) black-box modeling approach based on the cascade correlation (CC) algorithm formulated to calculate and predict multiplicity distribution of proton-proton (antiproton) (PP and PP ) inelastic interactions full phase space at a wide range of center-mass of energy . In addition, the formulated cascade correlation neural network (CCNN) model is used to empirically calculate the average multiplicity distribution nch> as a function of . The CCNN model was designed based on available experimental data for = 30.4 GeV, 44.5 GeV, 52.6 GeV, 62.2 GeV, 200 GeV, 300 GeV, 540 GeV, 900 GeV, 1000 GeV, 1800 GeV, and 7 TeV. Our obtained empirical results for P(nch), as well as nch> for (PP and PP) collisions are compared with the corresponding theoretical ones which obtained from other models. This comparison shows a good agreement with the available experimental data (up to 7 TeV) and other theoretical ones. At full large hadron collider (LHC) energy ( = 14 TeV) we have predicted P(nch) and nch> which also, show a good agreement with different theoretical models.

Black-box membership inference attacks based on shadow model

Membership inference attacks on machine learning models have drawn significant attention.While current research primarily utilizes shadow modeling techniques,which require knowledge of the target model and training data,practical scenarios involve black-box access to the target model with no available information.Limited training data further complicate the implementation of these attacks.In this paper,we experimentally compare common data enhancement schemes and propose a data synthesis framework based on the variational autoencoder generative adversarial network(VAE-GAN)to extend the training data for shadow models.Meanwhile,this paper proposes a shadow model training algorithm based on adversarial training to improve the shadow model's ability to mimic the predicted behavior of the target model when the target model's information is unknown.By conducting attack experiments on different models under the black-box access setting,this paper verifies the effectiveness of the VAE-GAN-based data synthesis framework for improving the accuracy of membership inference attack.Furthermore,we verify that the shadow model,trained by using the adversarial training approach,effectively improves the degree of mimicking the predicted behavior of the target model.Compared with existing research methods,the method proposed in this paper achieves a 2%improvement in attack accuracy and delivers better attack performance.

Probabilistic movement primitive based motion learning for a lower limb exoskeleton with black-box optimization

As a wearable robot,an exoskeleton provides a direct transfer of mechanical power to assist or augment the wearer’s movement with an anthropomorphic configuration.When an exoskeleton is used to facilitate the wearer’s movement,a motion generation process often plays an important role in high-level control.One of the main challenges in this area is to generate in real time a reference trajectory that is parallel with human intention and can adapt to different situations.In this paper,we first describe a novel motion modeling method based on probabilistic movement primitive(ProMP)for a lower limb exoskeleton,which is a new and powerful representative tool for generating motion trajectories.To adapt the trajectory to different situations when the exoskeleton is used by different wearers,we propose a novel motion learning scheme based on black-box optimization(BBO)PIBB combined with ProMP.The motion model is first learned by ProMP offline,which can generate reference trajectories for use by exoskeleton controllers online.PIBB is adopted to learn and update the model for online trajectory generation,which provides the capability of adaptation of the system and eliminates the effects of uncertainties.Simulations and experiments involving six subjects using the lower limb exoskeleton HEXO demonstrate the effectiveness of the proposed methods.

Optimization of operating conditions in the steam turbine blade cascade using the black-box method

Water droplets cause corrosion and erosion,condensation loss,and thermal efficiency reduction in low-pressure steam turbines.In this study,multi-objective optimization was carried out using the black-box method through the automatic linking of a genetic algorithm(GA)and a computational fluid dynamics(CFD)code to find the optimal values of two design variables(inlet stagnation temperature and cascade pressure ratio)to reduce wetness in the last stages of turbines.The wet steam flow numerical model was used to calculate the optimization parameters,including wetness fraction rate,mean droplet radius,erosion rate,condensation loss rate,kinetic energy rate,and mass flow rate.Examining the validation results showed a good agreement between the experimental data and the numerical outcomes.According to the optimization results,the inlet stagnation temperature and the cascade pressure ratio were proposed to be 388.67(K)and 0.55(-),respectively.In particular,the suggested optimaltemperature and pressure ratio improved the liquid mass fraction and mean droplet radius by about 32%and 29%,respectively.Also,in the identified optimal operating state,the ratios of erosion,condensation loss,and kinetic energy fell by 76%,32.7%,and 15.85%,respectively,while the mass flow rate ratio rose by 0.68%.

算法解释制度的体系化构建

算法解释在算法治理中举足轻重。算法解释承载着权益保障、社会交往和风险治理三重意义,其在技术层面上的障碍正在逐渐被突破,可以通过多种技术机制实现。在算法治理活动中,应根据常规场景、关键场景和争议场景分别限定选择解释的路径选择和技术方案,通过冻结机制、抽样机制和镜像机制固定算法解释,并使之接受外部的验证与审查,确保算法解释真实、有效。算法解释的系列机制应被进一步构建为体系化的算法解释制度,在这一制度框架内,对解释路径与精度、解释时限、解释瑕疵责任等要素的合理配置,可以实现社会效益与规制负担的精细平衡。

基于多能流耦合规律的综合能源系统潮流及流分析

从能量流和流两个角度分析综合能源系统的多能流耦合规律,建立系统能流稳态潮流模型和流模型,根据模型特点研究系统潮流和流分布计算方法,以算例验证计算方法的可行性并分析计算数据,验证流机理模型相比于黑箱模型在局部分析方面的优越性。以算例数据为例,分析其潮流及流分布特点,结果表明:算例系统中能源转换环节的地方损最大,为系统薄弱环节,值得着重改进;同时电、热网络也有部分管段损较大,优化局部网络时也应纳入考虑。

基于损失平滑的对抗样本攻击方法

深度神经网络(DNNs)容易受到对抗样本的攻击,现有基于动量的对抗样本生成方法虽然可以达到接近100%的白盒攻击成功率,但是在攻击其他模型时效果仍不理想,黑盒攻击成功率较低。针对此,提出一种基于损失平滑的对抗样本攻击方法来提高对抗样本的可迁移性。在每一步计算梯度的迭代过程中,不直接使用当前梯度,而是使用局部平均梯度来累积动量,以此来抑制损失函数曲面存在的局部振荡现象,从而稳定更新方向,逃离局部极值点。在ImageNet数据集上的大量实验结果表明:所提方法与现有基于动量的方法相比,在单个模型攻击实验中的平均黑盒攻击成功率分别提升了38.07%和27.77%,在集成模型攻击实验中的平均黑盒攻击成功率分别提升了32.50%和28.63%。

基于改进蚁群算法的对抗样本生成模型

传统的文本生成对抗方法主要采用位置置换、字符替换等方式,耗费时间较长且效果较差。针对以上问题,该文提出一种基于改进蚁群算法的对抗样本生成模型IGAS(Improved ant colony algorithm to Generate Adversarial Sample),利用蚁群算法的特点生成对抗样本,并利用类形字进行优化。首先,构建城市节点群,利用样本中的词构建城市节点群;然后对原始输入样本,利用改进的蚁群算法生成对抗样本;再针对生成结果,通过构建的中日类形字典进行字符替换,生成最终的对抗样本;最后在黑盒模式下进行对抗样本攻击实验。实验在情感分类、对话摘要生成、因果关系抽取等多种领域验证了该方法的有效性。

黑盒与逆向工程概念在数字电路基础实验教学中的应用

随着电子信息技术的飞速发展,传统数字电路实验内容虽较为经典,其授课模式难以真实反映学生实验完成情况,难以有效促进学生对理论课程知识的深入理解。对此摒弃传统FPGA黑盒实验系统中的正向验证方式,采取逆向分析方法,通过对电路外特性测试,进行电路功能判断,强化学生数字电路理论知识,提高学生逆向工程、自主设计和工程应用能力。实验设计符合工程教育认证标准,提高数字电路实验教学质量。

数字金融的算法风险及其法律规制

算法与金融的深度融合是金融产业数字化转型的关键驱动力,但同时也引发了算法歧视、算法权力、算法共谋、算法趋同等一系列新型风险。这些风险具有隐蔽性、系统性和根源多样性的特征,导致传统金融法制在法律主体、调整对象、算法权利、算法责任等方面表现出局限性。究其根源,在于算法黑箱削弱了金融消费者和监管者的风险识别和防范能力,并导致算法责任追究困难。因此,对数字金融算法风险的法律规制应以算法黑箱为切入点,以促进算法发展和防范算法风险统筹为目标,以算法透明性为基本原则,确保透明是合理的、有意义的且具有层次差异。在法律主体方面,应加强监管部门的专业性和协调机制,引入多方主体协同共治;在调整对象方面,应对算法本体和参与主体进行双维规制;在算法权利方面,应加强金融消费者数据保护与算法规制的衔接;在算法责任方面,应构建以算法设计与应用监管并重的双层监管结构,以确保数字金融算法安全可信的实现。

带隐藏约束昂贵黑箱问题的自适应代理优化方法

针对带隐藏约束的昂贵黑箱全局优化问题,提出采用自适应转换搜索策略的代理优化方法。在转换搜索子步中采用与已估值点个数相关的标准差在当前最优点附近通过随机扰动生成候选点,以更好地平衡局部搜索和全局搜索。为更好地近似真实黑箱目标函数,采用了自适应组合目标代理模型。在50个测试问题上进行了数值实验,计算结果说明了所提算法的有效性。

人工智能时代的算法黑箱与信任重建

信任是人机互动和人际互动的基石。然而,由于算法黑箱的不透明性和不可解释性,人工智能系统及其服务在赋权和便利一部分人的同时,算法滥用、算法误用和算法垄断也在侵蚀、排斥和边缘化另外一部分人的利益。就此而言,算法即重构、程序即政治,在智能化环境中,信任被破坏和侵蚀不仅存在于人机互动之间,人与人之间的传统信任关系也因在线匿名和算法驱动而陷入衰退。正是考虑到人工智能带来的种种社会信任挑战,以算法规制为核心的人工智能治理首先应着眼于以人为本构建可信赖人工智能生态体系,并基于技术秉性和道德法律框架重建人机信任关系和人际信任关系。

智媒时代互联网广告的算法“黑箱”与治理路径探析

随着互联网技术的更新迭代,人工智能、算法等技术逐渐嵌入日常随处可见的广告中,也因其“黑箱”问题给广告治理带来权力滥用、技术异化、隐私窥视等难题。算法“黑箱”的封闭性、歧视性和风险性等特征使算法在应用中直接影响互联网广告的制度。破解这些难题的路径在于:规制算法权力、改革算法生态、设立数据收集壁垒、推进“黑箱”透明。

“避男标签”:小红书女性用户的算法抵抗研究

互联网平台的发展使算法日益成为一种新型社会权力,影响着人们在数字社会的行为决策,也激发了人们对算法的警惕和抵抗。本文聚焦小红书平台的“避男标签”现象,探究小红书女性用户如何基于算法想象,使用特定的话题标签开展算法抵抗。研究发现,女性使用“避男标签”是为躲避网络社区的男性凝视和网络骚扰,在具体实践中主要通过多元化标签内容以实现多维度的算法抵抗。但“避男标签”的效果并不稳定,需要及时调整,开展动态的算法博弈。总体来看,女性通过创造性使用平台功能开展算法抵抗,是为了寻求性别平等、包容友好的网络空间。而解决问题的关键不仅仅在于个体的能动抵抗,还要积极推动网络空间治理。

一种基于后门技术的深度强化学习水印框架

深度强化学习(DRL)已经证明了它在各种复杂任务中的有效性,因其出色的性能使其商业化正在急剧加速。生成一个DRL模型需要大量的计算资源和专业知识,使得一个训练有素的DRL模型已经成为人工智能应用程序和产品的核心知识产权。基于对DRL模型的产权保护,防止非法抄袭、未经授权的分发和复制,提出一种后门技术的DRL水印框架DrlWF,并使用一个全新的评价指标水印动作实现比例来衡量水印性能。通过向训练状态中添加水印,并使用带有水印的水印状态训练模型从而实现将水印嵌入至模型中。框架中的水印嵌入操作可以通过将水印嵌入到少量的训练数据中(仅需0.025%的训练数据)和不影响性能的奖励修改来实现。实验结果证明,在标准状态下,DRL模型仍具有良好的性能;在水印状态下,DRL模型性能将急剧下降,不足原有性能的1%,且水印动作执行比例达到了99%。通过急剧下降的性能以及模型对水印状态的动作表现,即可验证模型的所有权。此外,该水印具有良好的鲁棒性,在模型微调和模型压缩下,模型依然能够识别出水印,性能急剧下降且水印动作执行比例依旧达到了99%以上,证明了该DRL水印具有良好的鲁棒性。

基于随机平滑的通用黑盒认证防御

近年来,基于深度神经网络(DNNs)的图像分类模型在人脸识别、自动驾驶等关键领域得到了广泛应用,并展现出卓越的性能.然而,深度神经网络容易受到对抗样本攻击,从而导致模型错误分类.为此,提升模型自身的鲁棒性已成为一个主要的研究方向.目前大部分的防御方法,特别是经验防御方法,都基于白盒假设,即防御者拥有模型的详细信息,如模型架构和参数等.然而,模型所有者基于隐私保护的考虑不愿意共享模型信息.即使现有的黑盒假设的防御方法,也无法防御所有范数扰动的攻击,缺乏通用性.因此,本文提出了一种适用于黑盒模型的通用认证防御方法.具体而言,本文首先设计了一个基于查询的无数据替代模型生成方案,在无需模型的训练数据与结构等先验知识的情况下,利用查询和零阶优化生成高质量的替代模型,将认证防御场景转化为白盒,确保模型的隐私安全.其次,本文提出了基于白盒替代模型的随机平滑和噪声选择方法,构建了一个能够抵御任意范数扰动攻击的通用认证防御方案.本文通过分析比较原模型和替代模型在白盒认证防御上的性能,确保了替代模型的有效性.相较于现有方法,本文提出的通用黑盒认证防御方案在CIFAR10数据集上的效果取得了显著的提升.实验结果表明,本文方案可以保持与白盒认证防御方法相似的效果.与之前基于黑盒的认证防御方法相比,本文方案在实现了所有L p的认证防御的同时,认证准确率提升了20%以上.此外,本文方案还能有效保护原始模型的隐私,与原始模型相比,本文方案使成员推理攻击的成功率下降了5.48%.

一种新型深度分类神经网络黑盒指纹水印算法

提出了一种新型的强鲁棒黑盒指纹水印框架及方法。首先,提出了一种基于数字水印技术的高视觉质量的、具有一定安全性的毒化图像构造方法,将指示用户身份的信息嵌入到毒化图像,实现多用户场景下深度神经网络模型的可追溯性,并降低毒化图像被伪造的概率;其次,提出了毒化特征加强模块来优化模型训练;最后,设计了对抗训练策略,有效地学习到嵌入强度很小的指纹水印。大量的仿真实验表明,所构造的毒化图像中的指纹水印具有非常好的隐蔽性,大幅超越了WaNet等同类最优模型水印方法;以分类性能降低不超过2.4%的代价获得了超过99%的黑盒模型指纹水印验证率;且即便在指纹水印相差1位时亦能准确地进行模型水印版权验证。这些性能总体上优于同类最优的模型水印方法,表明了所提方法的可行性和有效性。