• Publications
  • Influence
Towards Evaluating the Security of Real-World Deployed Image CAPTCHAs
TLDR
The first systematic study on the security of image captchas in the wild is conducted, identifying the design flaws of those popular schemes, the best practices, and the design principles towards more secureCaptchas.
Online E-Commerce Fraud: A Large-Scale Detection and Analysis
TLDR
Light is expected to shed light on defending against online frauds for practical e-commerce platforms by developing an efficient and scalable AnTi-Fraud system (ATF) and deploying it on the Taobao platform of Alibaba, which is one of the world's largest e- commerce platforms.
Privacy Leakage of Real-World Vertical Federated Learning
TLDR
This paper considers an honest-but-curious adversary who participants in training a distributed ML model, does not deviate from the defined learning protocol, but attempts to infer private training data from the legitimately received information.
Towards understanding the security of modern image captchas and underground captcha-solving services
TLDR
This paper proposes simple yet powerful attack frameworks against popular image captchas, identifies some design flaws in these popular schemes, and examines the underground market for captcha-solving services, identifying 152 such services.
CATS: Cross-Platform E-Commerce Fraud Detection
TLDR
This paper implements the design of CATS into a prototype system and evaluates this prototype on the world's popular e-commerce platform Taobao, showing that CATS can achieve a high accuracy of 91% in detecting frauds and conducts a comprehensive analysis on the reported frauds.
A Large-Scale Empirical Study on the Vulnerability of Deployed IoT Devices
TLDR
A ten-month-long empirical study on the vulnerability of 1,362,906 IoT devices varying from six types shows sufficient evidence that N-days vulnerability is seriously endangering the IoT devices.
Rare Category Detection Forest
TLDR
A novel tree-based algorithm known as RCD-Forest with time complexity and high query efficiency where n is the size of the unlabeled data set and the effectiveness and efficiency of this method is verified.
De-Health: All Your Online Health Information Are Belong to Us
  • S. Ji, Qinchen Gu, Ting Wang
  • Computer Science, Medicine
    IEEE 36th International Conference on Data…
  • 2 February 2019
TLDR
A novel online health data De-Anonymization (DA) framework, named De-Health, is presented and the DA efficacy is validated, and a linkage attack framework is presented which can link online health/medical information to real world people.
DeT: Defending Against Adversarial Examples via Decreasing Transferability
TLDR
DeT is a transferability-based defense method, which to the best of the knowledge is the first such attempt, which can defend against adversarialExamples generated by common attacks, and correctly label adversarial examples with both small and large perturbations.
FDI: Quantifying Feature-based Data Inferability
TLDR
This paper quantifies the conditions to have a desired fraction of the target users to be Top-K inferable and evaluates the user inferability in two cases: network traffic attribution in network forensics and feature-based data de-anonymization.
...
...