• Publications
  • Influence
LiveSky: Enhancing CDN with P2P
TLDR
We present the design and deployment experiences with LiveSky, a commercial hybrid CDN-P2P live streaming system, which inherits the best of both CDN and P2P. Expand
  • 58
  • 8
Unrestricted Adversarial Examples via Semantic Manipulation
TLDR
We introduce "unrestricted" perturbations that manipulate semantically meaningful image-based visual descriptors -- color and texture -- in order to generate effective and photorealistic adversarial examples. Expand
  • 23
  • 6
  • PDF
FuPlex: A full duplex MAC for the next generation WLAN
TLDR
A simple and compatible full duplex MAC protocol named FuPlex is proposed. Expand
  • 14
  • 3
CMFL: Mitigating Communication Overhead for Federated Learning
TLDR
Federated Learning enables mobile users to collaboratively learn a global prediction model by aggregating their individual updates without sharing the privacy-sensitive data. Expand
  • 28
  • 3
  • PDF
SHREC'13 Track: Large-Scale Partial Shape Retrieval Using Simulated Range Images
TLDR
We evaluate the performance of existing algorithms for partial shape retrieval using a large set of queries composed of views extracted from a 3D dataset. Expand
  • 18
  • 3
  • PDF
RAB: Provable Robustness Against Backdoor Attacks
TLDR
We propose the first robust training process, RAB, to certify model robustness against backdoor attacks, and propose an exact smooth-training algorithm for Gaussian and Uniform smoothing distributions. Expand
  • 8
  • 3
  • PDF
REFIT: a Unified Watermark Removal Framework for Deep Learning Systems with Limited Data
TLDR
We propose REFIT, a unified watermark removal framework based on fine-tuning, which does not rely on the knowledge of the watermarks and even the watermarking schemes. Expand
  • 7
  • 3
  • PDF
Gotta Catch'Em All: Using Honeypots to Catch Adversarial Attacks on Neural Networks
TLDR
We introduce trapdoors, honeypot weaknesses in the classification manifold that attract attackers searching for adversarial examples, leading them to produce attacks similar to trapdoors. Expand
  • 6
  • 2
  • PDF
AdvCodec: Towards A Unified Framework for Adversarial Text Generation
TLDR
We propose a general targeted attack framework AdvCodec for adversarial text generation which addresses the challenge of discrete input space and is easily adapted to general NLP tasks. Expand
  • 5
  • 2
QEBA: Query-Efficient Boundary-Based Blackbox Attack
TLDR
In this paper, we propose a Query-Efficient Boundary-based blackbox Attack (QEBA) based only on model’s final prediction labels. Expand
  • 6
  • 2
  • PDF