Search for: All records

When trained on biased datasets, Deep Neural Networks (DNNs) often make predictions based on attributes derived from features spuriously correlated with the target labels. This is especially problematic if these irrelevant features are easier for the model to learn than the truly relevant ones. Many existing approaches, called debiasing methods, have been proposed to address this issue, but they often require predefined bias labels and entail significantly increased computational complexity by incorporating extra auxiliary models. Instead, we provide an orthogonal perspective from the existing approaches, inspired by cognitive science, specifically Global Workspace Theory (GWT). Our method, Debiasing Global Workspace (DGW), is a novel debiasing framework that consists of specialized modules and a shared workspace, allowing for increased modularity and improved debiasing performance. Additionally, DGW enhances the transparency of decision-making processes by visualizing which features of the inputs the model focuses on during training and inference through attention masks. We begin by proposing an instantiation of GWT for the debiasing method. We then outline the implementation of each component within DGW. At the end, we validate our method across various biased datasets, proving its effectiveness in mitigating biases and improving model performance. 

Generative models have enabled the creation of contents that are indistinguishable from those taken from nature. Open-source development of such models raised concerns about the risks of their misuse for malicious purposes. One potential risk mitigation strategy is to attribute generative models via fingerprinting. Current fingerprinting methods exhibit a significant tradeoff between robust attribution accuracy and generation quality while lacking design principles to improve this tradeoff. This paper investigates the use of latent semantic dimensions as fingerprints, from where we can analyze the effects of design variables, including the choice of fingerprinting dimensions, strength, and capacity, on the accuracy-quality tradeoff. Compared with previous SOTA, our method requires minimum computation and is more applicable to large-scale models. We use StyleGAN2 and the latent diffusion model to demonstrate the efficacy of our method. 

Generative models are now capable of synthesizing images, speeches, and videos that are hardly distinguishable from authentic contents. Such capabilities cause concerns such as malicious impersonation and IP theft. This paper investigates a solution for model attribution, i.e., the classification of synthetic contents by their source models via watermarks embedded in the contents. Building on past success of model attribution in the image domain, we discuss algorithmic improvements for generating user-end speech models that empirically achieve high attribution accuracy, while maintaining high generation quality. We show the tradeoff between attributability and generation quality under a variety of attacks on generated speech signals attempting to remove the watermarks, and the feasibility of learning robust watermarks against these attacks. 

Growing applications of generative models have led to new threats such as malicious personation and digital copyright infringement. One solution to these threats is model attribution, i.e., the identification of user-end models where the contents under question are generated from. Existing studies showed empirical feasibility of attribution through a centralized classifier trained on all user-end models. However, this approach is not scalable in reality as the number of models ever grows. Neither does it provide an attributability guarantee. To this end, this paper studies decentralized attribution, which relies on binary classifiers associated with each user-end model. Each binary classifier is parameterized by a user-specific key and distinguishes its associated model distribution from the authentic data distribution. We develop sufficient conditions of the keys that guarantee an attributability lower bound. Our method is validated on MNIST, CelebA, and FFHQ datasets. We also examine the trade-off between generation quality and robustness of attribution against adversarial post-processes. 

The emergence of programmable switches offers a new opportunity to revisit ISP-scale defenses for volumetric DDoS attacks. In theory, these can offer better cost vs. performance vs. flexibility trade-offs relative to proprietary hardware and virtual appliances. However, the ISP setting creates unique challenges in this regard---we need to run a broad spectrum of detection and mitigation functions natively on the programmable switch hardware and respond to dynamic adaptive attacks at scale. Thus, prior efforts in using programmable switches that assume out-of-band detection and/or use switches merely as accelerators for specific tasks are no longer sufficient, and as such, this potential remains unrealized. To tackle these challenges, we design and implement Jaqen, a switch-native approach for volumetric DDoS defense that can run detection and mitigation functions entirely inline on switches, without relying on additional data plane hardware. We design switch-optimized, resource-efficient detection and mitigation building blocks. We design a flexible API to construct a wide spectrum of best-practice (and future) defense strategies that efficiently use switch capabilities. We build a network-wide resource manager that quickly adapts to the attack posture changes. Our experiments show that Jaqen is orders of magnitude more performant than existing systems: Jaqen can handle large-scale hybrid and dynamic attacks within seconds, and mitigate them effectively at high line-rates (380 Gbps). 

The emergence of programmable switches offers a new opportunity to revisit ISP-scale defenses for volumetric DDoS attacks. In theory, these can offer better cost vs. performance vs. flexibility trade-offs relative to proprietary hardware and virtual appliances. However, the ISP setting creates unique challenges in this regard---we need to run a broad spectrum of detection and mitigation functions natively on the programmable switch hardware and respond to dynamic adaptive attacks at scale. Thus, prior efforts in using programmable switches that assume out-of-band detection and/or use switches merely as accelerators for specific tasks are no longer sufficient, and as such, this potential remains unrealized. To tackle these challenges, we design and implement Jaqen, a switch-native approach for volumetric DDoS defense that can run detection and mitigation functions entirely inline on switches, without relying on additional data plane hardware. We design switch-optimized, resource-efficient detection and mitigation building blocks. We design a flexible API to construct a wide spectrum of best-practice (and future) defense strategies that efficiently use switch capabilities. We build a network-wide resource manager that quickly adapts to the attack posture changes. Our experiments show that Jaqen is orders of magnitude more performant than existing systems: Jaqen can handle large-scale hybrid and dynamic attacks within seconds, and mitigate them effectively at high line-rates (380 Gbps).