BloomDT-An improved privacy-preserving decision tree inference scheme

Outsourcing decision tree models to cloud servers can allow model providers to distribute their models at scale without purchasing dedicated hardware for model hosting.However,model providers may be forced to disclose private model details when hosting their models in the cloud.Due to the time and monetary investments associated with model training,model providers may be reluctant to host their models in the cloud due to these privacy concerns.Furthermore,clients may be reluctant to use these outsourced models because their private queries or their results may be disclosed to the cloud servers.In this paper,we propose BloomDT,a privacy-preserving scheme for decision tree inference,which uses Bloom filters to hide the original decision tree's structure,the threshold values of each node,and the order in which features are tested while maintaining reliable classification results that are secure even if the cloud servers collude.Our scheme's security and performance are verified through rigorous testing and analysis.

Data security and privacy computing in artificial intelligence

Artificial intelligence(AI)is a field of computer science dedicated to creating systems and algorithms that can perform tasks typically requiring human intelligence,such as learning,problem-solving,language understanding,and decision-making,contributing to a wide array of applications across diverse industries.The development of AI,such as machine learning and deep learning,has revolutionized data processing and analysis.This transformation is rapidly changing human life and has allowed for many practical AI based applications,including biometric recognition,text/sentimental analysis,and attack detection in the fields of health care,finance,autonomous vehicles,personalized recommendations.However,the potential benefits of AI are hindered by issues,such as insecurity and privacy violations in data processing and communication.

Energy-Theft Detection Issues for Advanced Metering Infrastructure in Smart Grid

With the proliferation of smart grid research, the Advanced Metering Infrastructure （AMI） has become the first ubiquitous and fixed computing platform. However, due to the unique characteristics of AMI, such as complex network structure, resource-constrained smart meter, and privacy-sensitive data, it is an especially challenging issue to make AMI secure. Energy theft is one of the most important concerns related to the smart grid implementation. It is estimated that utility companies lose more than S25 billion every year due to energy theft around the world. To address this challenge, in this paper, we discuss the background of AMI and identify major security requirements that AMI should meet. Specifically, an attack tree based threat model is first presented to illustrate the energy-theft behaviors in AMI. Then, we summarize the current AMI energy-theft detection schemes into three categories, i.e., classification-based, state estimation-based, and game theory-based ones, and make extensive comparisons and discussions on them. In order to provide a deep understanding of security vulnerabilities and solutions in AMI and shed light on future research directions, we also explore some open challenges and potential solutions for energy-theft detection.

Privacy-friendly spatial crowdsourcing in vehic-ular networks

With the evolution of conventional VANETs(Vehicle Ad-hoc Networks)into the IoV(Internet of Vehicles),vehicle-based spatial crowdsourcing has become a potential solution for crowdsourcing applications.In vehicular networks,a spatial-temporal task/question can be outsourced(i.e.,task/question relating to a particular location and in a speci c time period)to some suitable smart vehicles(also known as workers)and then these workers can help solve the task/question.However,an inevitable barrier to the widespread deployment of spatial crowdsourcing applications in vehicular networks is the concern of privacy.Hence,We propose a novel privacy-friendly spatial crowdsourcing scheme.Unlike the existing schemes,the proposed scheme considers the privacy issue from a new perspective according that the spatial-temporal tasks can be linked and analyzed to break the location privacy of workers.Speci cally,to address the challenge,three privacy requirements(i.e.anonymity,untraceability,and unlinkability)are de ned and the proposed scheme combines an effcient anonymous technique with a new composite privacy metric to protect against attackers.Detailed privacy analyses show that the proposed scheme is privacy-friendly.In addition,performance evaluations via extensive simulations are also conducted,and the results demonstrate the effciency and e ectiveness of the proposed scheme.