基于联邦学习的本地化差分隐私机制研究

联邦学习与群体学习作为当前热门的分布式机器学习范式,前者能够保护用户数据不被第三方获得的前提下在服务器中实现模型参数共享计算,后者在无中心服务器的前提下利用区块链技术实现所有用户同等地聚合模型参数。但是,通过分析模型训练后的参数,如深度神经网络训练的权值,仍然可能泄露用户的隐私信息。目前,在联邦学习下运用本地化差分隐私(LDP)保护模型参数的方法层出不穷,但皆难以在较小的隐私预算和用户数量下缩小模型测试精度差。针对此问题,该文提出正负分段机制(PNPM),在聚合前对本地模型参数进行扰动。首先,证明了该机制满足严格的差分隐私定义,保证了算法的隐私性;其次分析了该机制能够在较少的用户数量下保证模型的精度,保证了机制的有效性;最后,在3种主流图像分类数据集上与其他最先进的方法在模型准确性、隐私保护方面进行了比较,表现出了较好的性能。     

Safeguarding cross-silo federated learning with local differential privacy

Federated Learning (FL) is a new computing paradigm in privacy-preserving Machine Learning (ML), where the ML model is trained in a decentralized manner by the clients, preventing the server from directly accessing privacy-sensitive data from the clients. Unfortunately, recent advances have shown potential risks for user-level privacy breaches under the cross-silo FL framework. In this paper, we propose addressing the issue by using a three-plane framework to secure the cross-silo FL, taking advantage of the Local Differential Privacy (LDP) mechanism. The key insight here is that LDP can provide strong data privacy protection while still retaining user data statistics to preserve its high utility. Experimental results on three real-world datasets demonstrate the effectiveness of our framework.     

Enhancing IoT anomaly detection performance for federated learning

Federated Learning (FL) with mobile computing and the Internet of Things (IoT) is an effective cooperative learning approach. However, several technical challenges still need to be addressed. For instance, dividing the training process among several devices may impact the performance of Machine Learning (ML) algorithms, often significantly degrading prediction accuracy compared to centralized learning. One of the primary reasons for such performance degradation is that each device can access only a small fraction of data (that it generates), which limits the efficacy of the local ML model constructed on that device. The performance degradation could be exacerbated when the participating devices produce different classes of events, which is known as the class balance problem. Moreover, if the participating devices are of different types, each device may never observe the same types of events, which leads to the device heterogeneity problem. In this study, we investigate how data augmentation can be applied to address these challenges and improving detection performance in an anomaly detection task using IoT datasets. Our extensive experimental results with three publicly accessible IoT datasets show the performance improvement of up to 22.9% with the approach of data augmentation, compared to the baseline (without relying on data augmentation). In particular, stratified random sampling and uniform random sampling show the best improvement in detection performance with only a modest increase in computation time, whereas the data augmentation scheme using Generative Adversarial Networks is the most time-consuming with limited performance benefits.     

基于相似度聚类的可信联邦安全聚合算法

联邦学习能够有效地规避参与方数据隐私问题,但模型训练中传递的参数或者梯度仍有可能泄露参与方的隐私数据,而恶意参与方的存在则会严重影响聚合过程和模型质量。基于此,该文提出一种基于相似度聚类的可信联邦安全聚合方法(FSA-SC)。首先基于客户端训练数据集规模及其与服务器间的通信距离综合评估选出拟参与模型聚合的候选客户端;然后根据候选客户端间的相似度,利用聚类将候选客户端划分为良性客户端和异常客户端;最后,对异常客户端类中的成员利用类内广播和二次协商进行参数替换和记录,检测识别恶意客户端。为了验证FSA-SC的有效性,以联邦推荐为应用场景,选取MovieLens 1M,Netflix数据集和Amazon抽样数据集为实验数据集,实验结果表明,所提方法能够实现高效的安全聚合,且相较对比方法有更高的鲁棒性。     

Privacy-preserving and accountable on-the-road prosecution of invalid vehicular mandatory authorizations

Nowadays, improving road safety is one of the major challenges in developed countries and, to this regard, attaining more effectiveness in the enforcement of road safety policies has become a key target. In particular, enforcing the requirements related to the technical and administrative mandatory documentation of on-the-road motor vehicles is one of the critical issues. The use of modern technologies in the context of Intelligent Transportation Systems (ITS) could enable the design of a more convenient, frequent and effective enforcement system compared to the traditional human patrol controls. In this article we propose a novel system for the on-the-fly verification of mandatory technical and administrative documentation of motor vehicles. Vehicles not complying with the required regulations will be identified and sanctioned whereas those vehicles, observant of the mandatory regulations, will maintain anonymity and non-traceability of their whereabouts. The proposed system is based on the use of anonymous credentials which will be loaded onto the vehicle to automatically and on-the-fly prove holdership of required credentials without requiring the vehicle to stop beside the road. We also implement a prototype of the credential system and analyze the feasibility of our solution in terms of computational cost and time to perform such telematic controls.     

基于SNA的群体推荐系统的研究

由于信息爆炸问题,如何为用户提供有效的个性化信息服务已得到广泛关注,而随着社交网络的流行及其带来的大量网络群体,如何为群体提供更好的个性化推荐服务也变得越来越重要.文中不仅考虑用户的兴趣偏好,同时利用社会网络分析法（Social Network Analysis,简称SNA）衡量用户之间的社会关系,将此因素融入推荐过程,实验证明此方法能够取得较好的推荐效果.     

基于k-匿名的隐私保护计算卸载方法

针对移动边缘计算(MEC)中用户的卸载任务及卸载频率可能使用户被攻击者锁定的问题,该文提出一种基于k-匿名的隐私保护计算卸载方法。首先,该方法基于用户间卸载任务及其卸载频率的差异性,提出隐私约束并建立基于卸载频率的隐私保护计算卸载模型;然后,提出基于模拟退火的隐私保护计算卸载算法(PCOSA)求得最优的k-匿名分组结果和组内各任务的隐私约束频率;最后,在卸载过程中改变用户原始卸载频率满足隐私约束,最小化终端能耗。仿真结果表明,PCOSA算法能找出用户所处MEC节点下与用户卸载表现最相近的k个用户形成匿名集,有效保护了所有用户隐私。     

基于改进的协同过滤图书推荐算法

高校图书馆图书个性化推荐没有得到很好的推广和实施,一个重要原因是用户对图书的评价不足.因此,提出了一种基于兴趣的高校图书推荐算法.该算法较好地解决了协同过滤算法无法使用和评分不足的问题.同时,将流行与反向流行的特征结合,使其更接近读者的行为.实验表明,该算法优于传统的协同过滤推荐算法,能够满足实际需求.     

高效的隐私保护在线开票服务认证方案

近年来,发票形式由传统的纸质凭据向电子凭据转变。相比于开具纸质凭据,在线开具电子凭据具有流程简化、成本降低以及便于存储等优势。但是,如何保证在线开具电子凭据服务中实体身份的合法性以及身份信息的隐私性是当前研究的重点问题。为了解决此问题,利用预共享密钥机制,该文提出一种隐私保护在线开具电子凭据的认证方案。在此方案中,合法用户与企业完成交易后可以本地在线发起开票申请,国家税务总局的电子凭据系统成功核验实体身份和交易信息后可为该用户提供电子凭据。安全和性能分析结果表明提出方案可以在耗费较少认证开销的情况下提供鲁棒的安全属性。     

基于HBase的实时消息推荐系统

在现有的推荐系统中,基于用户兴趣模型都能够表达出用户的兴趣,但在用户兴趣发生变化时却不能够及时更新模型。提出基于用户反馈内容来实时更新用户兴趣的消息推荐系统,通过实时更新模型和特征向量进而得到用户当前最匹配的推荐结果。并使用HBase(Hadoop Database)作为存储,能更好地适应数据规模的增长。     

BPR-UserRec:a personalized user recommendation method in social tagging systems

Social tagging is one of the most important characteristics of Web 2.0 services, and social tagging systems (STS) are becoming more and more popular for users to annotate, organize and share items on the Web. Moreover, online social network has been incorporated into social tagging systems. As more and more users tend to interact with real friends on the Web, personalized user recommendation service provided in social tagging systems is very appealing. In this paper, we propose a personalized user recommendation method, and our method handles not only the users’ interest networks, but also the social network information. We empirically show that our method outperforms a state-of-the-art method on real dataset from Last.fm dataset and Douban.     

基于同态加密的可验证隐私保护联邦学习方案

现有基于同态加密的联邦学习安全和隐私保护方案中,仍面临着服务器伪造聚合结果或与用户合谋导致隐私数据泄露风险。针对上述问题,该文提出抗合谋的隐私保护和可验证联邦学习方案。首先,通过结合秘密共享算法实现密钥的生成和协作解密,并采用同态加密等密码学原语进一步保护模型,防止用户与服务器的合谋攻击。然后基于双线性聚合签名算法使每个用户能够独立验证服务器提供的聚合结果。同时,为了鼓励更多拥有高质量数据的用户参与进来,该文提出一种激励机制,为用户提供相应的奖励。安全性分析表明,该文方案对系统中存在的合谋攻击具有鲁棒性。最后,理论分析和实验验证结果表明该方案具有可靠性、可行性和有效性。     

Spark平台下教育资源个性化推荐研究

在面对海量教育数据处理情况时,传统的协同过滤算法在单机上训练和测试效率低下,针对该问题,提出了基于Hadoop分布式平台和Spark并行计算模型的无中间结果输出改进型教育资源推荐策略,该策略较好地发挥了Spark的迭代计算能力优势,在应用于教育资源推荐时,比较了传统算法与改进算法在分布式情况和非分布式情况下的推荐效率和推荐质量的情况.实验结果表明,利用Spark计算模型实现协同过滤算法能够有效地提高教育资源个性化推荐的推荐质量以及推荐效率.     

Activity event recommendation and attendance prediction

ABSTRACT

The recommendation problem has been widely studied and researchers are constantly searching for better methods. Recommending events is an even more difficult problem because there is no information such as ratings from past events. In this paper, we introduce a method for recommending activity events: activities hosted by one or more individuals which involve movement: walking, running, cycling, cross-country skiing, and driving to users who have location history such as trajectories, meetings, POI visits, and geo-tagged photos. We tested the method in a real environment in Mopsi platform: http://cs.uef.fi/mopsi/events. Although there are many location-based event recommendation systems in literature, this is to our knowledge the first system that recommends activity events like bicycle and skiing trips. The experiments show that we can predict whether a user is attending the event or not with 80% accuracy, which is significantly better than random chance (50%).     

个性化特征的电子商务智能推荐系统

以准确向用户推荐商品,提升电子商务网站销售量为目标,设计基于个性化特征的电子商务智能推荐系统.系统以个性化推荐引擎为核心,采集交易事务、商品特征、用户评价等数据,利用基于个性化特征的协同过滤推荐算法计算商品间相似度,确定新商品的近邻,根据近邻用户对新商品的评价结果选择商品进行推荐.测试结果表明,该系统的电子商务商品推荐...     

Gbit收发器的特点及PCB制板建议

随着软件无线电平台的提出和高速实时信号处理的不断发展,信号传输的速度越来越高,容量越来越大,可靠的板间通信业变得越来越困难了。而Gbit收发器的出现很好地解决了这个问题。本文介绍了Gbit收发器的特点,并针对其PCB制板比较复杂的现实情况,提出了PCB制板的相关建议。