Publications

To appear in proceedings of the 41st IEEE Symposium on Security and Privacy (S&P 2020)
This paper looks at past research conducted in the area of cyber insurance and classifies previous studies in four different areas. Then it identifies, a group of practical research problems where security experts could help the cyber insurance domain.

To appear in the proceedings of the Annual Conference of the PHM Society (PHM 2020)
We introduce turbidity detection as a practical super set of the adversarial input detection problem,coping with adversarial campaigns rather than statistically invisible one-offs. This perspective is coupled with ROC-theoretic design guidance that prescribes an inexpensive do-main adaptation layer at the output of a deep learning model during an attack campaign. The result aims to approximate the Bayes optimal mitigation that ameliorates the detection models degraded health.

To appear in the proceedings of ACM CHI Conference on Human Factors in Computing Systems (CHI 2020) (Honorable Mention Award)
Our online survey of 902 individuals studies the reasons for which users struggle to adhere to expert-recommended security, privacy, and identity-protection practices. We examined 30 of these practices, finding that gender, education, technical background, and prior negative experiences correlate with practice adoption levels. We found that practices were abandoned when they were perceived as low-value, inconvenient, or when overridden by subjective judgment. We discuss how tools and expert recommendations can better align to user needs.

To appear in proceedings of the 34th AAAI Conference on Artificial Intelligence (AAAI 2020)
For federated learning systems deployed in the wild, data flaws hosted on local agents are widely witnessed. On one hand, given a large amount of training data are corrupted by systematic sensor noise and environmental perturbations, the performances of federated model training can be degraded significantly. On the other hand, it is prohibitively expensive for either clients or service providers to set up manual sanitary checks to verify the quality of data instances. In our study, we echo this challenge by proposing a collaborative and privacy-preserving machine teaching method. Specifically, we use a few trusted instances provided by teachers as benign examples in the teaching process. Our collaborative teaching approach seeks jointly the optimal tuning on the distributed training set, such that the model learned from the tuned training set predicts labels of the trusted items correctly. The proposed method couples the process of teaching and learning and thus produces directly a robust prediction model despite the extremely pervasive systematic data corruption. The experimental study on real benchmark data sets demonstrates the validity of our method.

To appear in the proceedings of the 2020 USENIX Annual Technical Conference (USENIX ATC '20)
We propose SEAL, a family of new searchable encryption schemes with adjustable leakage. In SEAL, the amount of privacy loss is expressed in leaked bits of search or access pattern and can be defined at setup. As our experiments show, when protecting only a few bits of leakage (e.g., three to four bits of access pattern),enough for existing and even new more aggressive attacks to fail, SEAL query execution time is within the realm of practical for real-world applications (a little over one order of magnitude slowdown compared to traditional SE-based encrypted databases).