We are witnessing a confluence between applied cryptography and secure hardware systems in enabling cloud computing.On one hand, work has enabled efficient, oblivious data-structures memory primitives.On the other, emergence of Intel SGX low-overhead mass market mechanism for isolated execution.By themselves these technologies have their disadvantages.Oblivious primitives carry high performance overheads, especially when run non-interactively.Intel SGX, while more suffers from numerous...

Service robots are becoming a widespread tool for assisting humans in scientific, industrial and even domestic settings. Yet, our understanding of how to motivate sustain interactions between human users remains limited. In this work, we conducted study investigate surprising robot behaviour evokes curiosity influences trust engagement the context participants interacting with Recyclo, service providing recycling recommendations. Wizard-of-Oz experiment, 36 were asked interact Recyclo...

Abstract Anonymous communications networks enable individuals to maintain their privacy online. The most popular such network is Tor, with about two million daily users; however, Tor reaching limits of its scalability. One the main scalability bottlenecks and similar designs originates from requirement distributing a global view servers in all clients. This place avoid epistemic attacks , which adversaries who know parts certain clients do not can rule or out those being responsible for...

Hyperparameter optimization is a ubiquitous challenge in machine learning, and the performance of trained model depends crucially upon their effective selection. While rich set tools exist for this purpose, there are currently no practical hyperparameter selection methods under constraint differential privacy (DP). We study honest differentially private which process tuning accounted overall budget. To end, we i) show that standard composition outperform more advanced techniques many...

Protecting metadata of communications has been an area active research since the dining cryptographers problem was introduced by David Chaum in 1988. The Snowden revelations from 2013 resparked this direction. Consequently over last decade we have witnessed a flurry novel systems designed to protect users' online. However, such leverage different assumptions and design choices achieve their goal; resulting scattered view desirable properties, potential vulnerabilities, limitations existing...

Several privacy-preserving analytics frameworks have been proposed that use trusted execution environments (TEEs) like Intel SGX. Such often compaction and shuffling as core primitives. However, due to advances in TEE side-channel attacks, these primitives, the applications them, should be fully oblivious; is, perform instruction sequences memory accesses do not depend on secret inputs. obliviousness would eliminate threat of leaking private information through or timing side channels, but...

As more privacy-preserving solutions leverage trusted execution environments (TEEs) like Intel SGX, it becomes pertinent that these can by design thwart TEE side-channel attacks research has brought to light. In particular, such need be fully oblivious circumvent leaking private information through memory or timing side channels.

We present Private Random Access Computations (PRAC), a 3-party Secure Multi-Party Computation (MPC) framework to support random-access data structure algorithms for MPC with efficient communication in terms of rounds and bandwidth. PRAC extends the state-of-the-art DORAM Duoram new implementation, more flexibility how memory is shared, Incremental Wide DPFs. then use these DPF extensions achieve algorithmic improvements three novel oblivious protocols MPC. exploits observation that secure...

We study secure and privacy-preserving data analysis based on queries executed samples from a dataset. Trusted execution environments (TEEs) can be used to protect the content of during query computation, while supporting differential-private (DP) in TEEs provides record privacy when output is revealed. Support for sample-based attractive due \emph{privacy amplification} since not all dataset answer but only small subset. However, extracting with proving strong DP guarantees trivial as...

Hyperparameter optimization is a ubiquitous challenge in machine learning, and the performance of trained model depends crucially upon their effective selection. While rich set tools exist for this purpose, there are currently no practical hyperparameter selection methods under constraint differential privacy (DP). We study honest differentially private which process tuning accounted overall budget. To end, we i) show that standard composition outperform more advanced techniques many...