arXiv subjects

Shan-Hung Wu

Publications and source records attributed to Shan-Hung Wu.

3 recordsLinked to original sources

arXiv2022

Single-level Adversarial Data Synthesis based on Neural Tangent Kernels

Abstract Generative adversarial networks (GANs) have achieved impressive performance in data synthesis and have driven the development of many applications. However, GANs are known to be hard to train due to their bilevel objective, which leads to the problems of convergence, mode collapse, and gradient vanishing. In this paper, we propose a new generative model called the generative adversarial NTK (GA-NTK) that has a single-level objective. The GA-NTK keeps the spirit of adversarial learning (which helps generate plausible data) while avoiding the training difficulties of GANs. This is done by modeling the discriminator as a Gaussian process with a neural tangent kernel (NTK-GP) whose training dynamics can be completely described by a closed-form formula. We analyze the convergence behavior of GA-NTK trained by gradient descent and give some sufficient conditions for convergence. We also conduct extensive experiments to study the advantages and limitations of GA-NTK and propose some techniques that make GA-NTK more practical.

cs.LG↗

arXiv2019

Ego-CNN: Distributed, Egocentric Representations of Graphs for Detecting Critical Structures

We study the problem of detecting critical structures using a graph embedding model. Existing graph embedding models lack the ability to precisely detect critical structures that are specific to a task at the global scale. In this paper, we propose a novel graph embedding model, called the Ego-CNNs, that employs the ego-convolutions convolutions at each layer and stacks up layers using an ego-centric way to detects precise critical structures efficiently. An Ego-CNN can be jointly trained with a task model and help explain/discover knowledge for the task. We conduct extensive experiments and the results show that Ego-CNNs (1) can lead to comparable task performance as the state-of-the-art graph embedding models, (2) works nicely with CNN visualization techniques to illustrate the detected structures, and (3) is efficient and can incorporate with scale-free priors, which commonly occurs in social network datasets, to further improve the training efficiency.

cs.LG↗

arXiv2013

Is Somebody Watching Your Facebook Newsfeed?

With the popularity of Social Networking Services (SNS), more and more sensitive information are stored online and associated with SNS accounts. The obvious value of SNS accounts motivates the usage stealing problem -- unauthorized, stealthy use of SNS accounts on the devices owned/used by account owners without any technology hacks. For example, anxious parents may use their kids' SNS accounts to inspect the kids' social status; husbands/wives may use their spouses' SNS accounts to spot possible affairs. Usage stealing could happen anywhere in any form, and seriously invades the privacy of account owners. However, there is no any currently known defense against such usage stealing. To an SNS operator (e.g., Facebook Inc.), usage stealing is hard to detect using traditional methods because such attackers come from the same IP addresses/devices, use the same credentials, and share the same accounts as the owners do. In this paper, we propose a novel continuous authentication approach that analyzes user browsing behavior to detect SNS usage stealing incidents. We use Facebook as a case study and show that it is possible to detect such incidents by analyzing SNS browsing behavior. Our experiment results show that our proposal can achieve higher than 80% detection accuracy within 2 minutes, and higher than 90% detection accuracy after 7 minutes of observation time.

cs.SI↗