Search

In recent years, Text-to-Image (T2I) models have garnered significant attention due to their remarkable advancements. However, security concerns have emerged due to their potential to generate inappropriate or Not-Safe-For-Work (NSFW) images. In this paper, inspired by the observation that texts with different semantics can lead to similar human perceptions, we propose an LLM-driven perception-guided jailbreak method, termed PGJ. It is a black-box jailbreak method that requires no specific T2I model (model-free) and generates highly natural attack prompts. Specifically, we propose identifying a safe phrase that is similar in human perception yet inconsistent in text semantics with the target unsafe word and using it as a substitution. The experiments conducted on six open-source models and commercial online services with thousands of prompts have verified the effectiveness of PGJ., Comment: 8 pages

The widespread use of diffusion methods enables the creation of highly realistic images on demand, thereby posing significant risks to the integrity and safety of online information and highlighting the necessity of DeepFake detection. Our analysis of features extracted by traditional image encoders reveals that both low-level and high-level features offer distinct advantages in identifying DeepFake images produced by various diffusion methods. Inspired by this finding, we aim to develop an effective representation that captures both low-level and high-level features to detect diffusion-based DeepFakes. To address the problem, we propose a text modality-oriented feature extraction method, termed TOFE. Specifically, for a given target image, the representation we discovered is a corresponding text embedding that can guide the generation of the target image with a specific text-to-image model. Experiments conducted across ten diffusion types demonstrate the efficacy of our proposed method.

In the last few years, generative models have shown their powerful capabilities in synthesizing realistic images in both quality and diversity (i.e., facial images, and natural subjects). Unfortunately, the artifact patterns in fake images synthesized by different generative models are inconsistent, leading to the failure of previous research that relied on spotting subtle differences between real and fake. In our preliminary experiments, we find that the artifacts in fake images always change with the development of the generative model, while natural images exhibit stable statistical properties. In this paper, we employ natural traces shared only by real images as an additional predictive target in the detector. Specifically, the natural traces are learned from the wild real images and we introduce extended supervised contrastive learning to bring them closer to real images and further away from fake ones. This motivates the detector to make decisions based on the proximity of images to the natural traces. To conduct a comprehensive experiment, we built a high-quality and diverse dataset that includes generative models comprising 6 GAN and 6 diffusion models, to evaluate the effectiveness in generalizing unknown forgery techniques and robustness in surviving different transformations. Experimental results show that our proposed method gives 96.1% mAP significantly outperforms the baselines. Extensive experiments conducted on the widely recognized platform Midjourney reveal that our proposed method achieves an accuracy exceeding 78.4%, underscoring its practicality for real-world application deployment. The source code and partial self-built dataset are available in supplementary material.

In recent years, DeepFake technology has achieved unprecedented success in high-quality video synthesis, whereas these methods also pose potential and severe security threats to humanity. DeepFake can be bifurcated into entertainment applications like face swapping and illicit uses such as lip-syncing fraud. However, lip-forgery videos, which neither change identity nor have discernible visual artifacts, present a formidable challenge to existing DeepFake detection methods. Our preliminary experiments have shown that the effectiveness of the existing methods often drastically decreases or even fails when tackling lip-syncing videos. In this paper, for the first time, we propose a novel approach dedicated to lip-forgery identification that exploits the inconsistency between lip movements and audio signals. We also mimic human natural cognition by capturing subtle biological links between lips and head regions to boost accuracy. To better illustrate the effectiveness and advances of our proposed method, we curate a high-quality LipSync dataset by employing the SOTA lip generator. We hope this high-quality and diverse dataset could be well served the further research on this challenging and interesting field. Experimental results show that our approach gives an average accuracy of more than 95.3% in spotting lip-syncing videos, significantly outperforming the baselines. Extensive experiments demonstrate the capability to tackle deepfakes and the robustness in surviving diverse input transformations. Our method achieves an accuracy of up to 90.2% in real-world scenarios (e.g., WeChat video call) and shows its powerful capabilities in real scenario deployment. To facilitate the progress of this research community, we release all resources at https://github.com/AaronComo/LipFD., Comment: The first two authors contributed equally to this work

RAW to sRGB mapping, which aims to convert RAW images from smartphones into RGB form equivalent to that of Digital Single-Lens Reflex (DSLR) cameras, has become an important area of research. However, current methods often ignore the difference between cell phone RAW images and DSLR camera RGB images, a difference that goes beyond the color matrix and extends to spatial structure due to resolution variations. Recent methods directly rebuild color mapping and spatial structure via shared deep representation, limiting optimal performance. Inspired by Image Signal Processing (ISP) pipeline, which distinguishes image restoration and enhancement, we present a novel Neural ISP framework, named FourierISP. This approach breaks the image down into style and structure within the frequency domain, allowing for independent optimization. FourierISP is comprised of three subnetworks: Phase Enhance Subnet for structural refinement, Amplitude Refine Subnet for color learning, and Color Adaptation Subnet for blending them in a smooth manner. This approach sharpens both color and structure, and extensive evaluations across varied datasets confirm that our approach realizes state-of-the-art results. Code will be available at ~\url{https://github.com/alexhe101/FourierISP}.

Within the realm of computer vision, self-supervised learning (SSL) pertains to training pre-trained image encoders utilizing a substantial quantity of unlabeled images. Pre-trained image encoders can serve as feature extractors, facilitating the construction of downstream classifiers for various tasks. However, the use of SSL has led to an increase in security research related to various backdoor attacks. Currently, the trigger patterns used in backdoor attacks on SSL are mostly visible or static (sample-agnostic), making backdoors less covert and significantly affecting the attack performance. In this work, we propose GhostEncoder, the first dynamic invisible backdoor attack on SSL. Unlike existing backdoor attacks on SSL, which use visible or static trigger patterns, GhostEncoder utilizes image steganography techniques to encode hidden information into benign images and generate backdoor samples. We then fine-tune the pre-trained image encoder on a manipulation dataset to inject the backdoor, enabling downstream classifiers built upon the backdoored encoder to inherit the backdoor behavior for target downstream tasks. We evaluate GhostEncoder on three downstream tasks and results demonstrate that GhostEncoder provides practical stealthiness on images and deceives the victim model with a high attack success rate without compromising its utility. Furthermore, GhostEncoder withstands state-of-the-art defenses, including STRIP, STRIP-Cl, and SSL-Cleanse., Comment: 24 pages,8 figures

Self-supervised learning (SSL), a paradigm harnessing unlabeled datasets to train robust encoders, has recently witnessed substantial success. These encoders serve as pivotal feature extractors for downstream tasks, demanding significant computational resources. Nevertheless, recent studies have shed light on vulnerabilities in pre-trained encoders, including backdoor and adversarial threats. Safeguarding the intellectual property of encoder trainers and ensuring the trustworthiness of deployed encoders pose notable challenges in SSL. To bridge these gaps, we introduce SSL-Auth, the first authentication framework designed explicitly for pre-trained encoders. SSL-Auth leverages selected key samples and employs a well-trained generative network to reconstruct watermark information, thus affirming the integrity of the encoder without compromising its performance. By comparing the reconstruction outcomes of the key samples, we can identify any malicious alterations. Comprehensive evaluations conducted on a range of encoders and diverse downstream tasks demonstrate the effectiveness of our proposed SSL-Auth.

Adversarial training (AT) is widely considered the state-of-the-art technique for improving the robustness of deep neural networks (DNNs) against adversarial examples (AE). Nevertheless, recent studies have revealed that adversarially trained models are prone to unfairness problems, restricting their applicability. In this paper, we empirically observe that this limitation may be attributed to serious adversarial confidence overfitting, i.e., certain adversarial examples with overconfidence. To alleviate this problem, we propose HAM, a straightforward yet effective framework via adaptive Hard Adversarial example Mining.HAM concentrates on mining hard adversarial examples while discarding the easy ones in an adaptive fashion. Specifically, HAM identifies hard AEs in terms of their step sizes needed to cross the decision boundary when calculating loss value. Besides, an early-dropping mechanism is incorporated to discard the easy examples at the initial stages of AE generation, resulting in efficient AT. Extensive experimental results on CIFAR-10, SVHN, and Imagenette demonstrate that HAM achieves significant improvement in robust fairness while reducing computational cost compared to several state-of-the-art adversarial training methods. The code will be made publicly available.

In recent decades, Generative Adversarial Network (GAN) and its variants have achieved unprecedented success in image synthesis. However, well-trained GANs are under the threat of illegal steal or leakage. The prior studies on remote ownership verification assume a black-box setting where the defender can query the suspicious model with specific inputs, which we identify is not enough for generation tasks. To this end, in this paper, we propose a novel IP protection scheme for GANs where ownership verification can be done by checking outputs only, without choosing the inputs (i.e., box-free setting). Specifically, we make use of the unexploited potential of the discriminator to learn a hypersphere that captures the unique distribution learned by the paired generator. Extensive evaluations on two popular GAN tasks and more than 10 GAN architectures demonstrate our proposed scheme to effectively verify the ownership. Our proposed scheme shown to be immune to popular input-based removal attacks and robust against other existing attacks. The source code and models are available at https://github.com/AbstractTeen/gan_ownership_verification, Comment: Accepted to ICCV 2023. The first two authors contributed equally to this work

Many problems in database systems, such as cardinality estimation, database testing and optimizer tuning, require a large query load as data. However, it is often difficult to obtain a large number of real queries from users due to user privacy restrictions or low frequency of database access. Query generation is one of the approaches to solve this problem. Existing query generation methods, such as random generation and template-based generation, do not consider the relationship between the generated queries and existing queries, or even generate semantically incorrect queries. In this paper, we propose a query generation framework based on generative adversarial networks (GAN) to generate query load that is similar to the given query load. In our framework, we use a syntax parser to transform the query into a parse tree and traverse the tree to obtain the sequence of production rules corresponding to the query. The generator of GAN takes a fixed distribution prior as input and outputs the query sequence, and the discriminator takes the real query and the fake query generated by the generator as input and outputs a gradient to guide the generator learning. In addition, we add context-free grammar and semantic rules to the generation process, which ensures that the generated queries are syntactically and semantically correct. We conduct experiments to evaluate our approach on real-world dataset, which show that our approach can generate new query loads with a similar distribution to a given query load, and that the generated queries are syntactically correct with no semantic errors. The generated query loads are used in downstream task, and the results show a significant improvement in the models trained with the expanded query loads using our approach.

Watermarking has been widely adopted for protecting the intellectual property (IP) of Deep Neural Networks (DNN) to defend the unauthorized distribution. Unfortunately, the popular data-poisoning DNN watermarking scheme relies on target model fine-tuning to embed watermarks, which limits its practical applications in tackling real-world tasks. Specifically, the learning of watermarks via tedious model fine-tuning on a poisoned dataset (carefully-crafted sample-label pairs) is not efficient in tackling the tasks on challenging datasets and production-level DNN model protection. To address the aforementioned limitations, in this paper, we propose a plug-and-play watermarking scheme for DNN models by injecting an independent proprietary model into the target model to serve the watermark embedding and ownership verification. In contrast to the prior studies, our proposed method by incorporating a proprietary model is free of target model fine-tuning without involving any parameters update of the target model, thus the fidelity is well preserved. Our research findings reveal that model fine-tuning with poisoned data is not prepared for the IP protection of DNN models deployed in real-world tasks and poses a new research direction toward a more thorough understanding and investigation of adopting the proprietary model for DNN watermarking. The source code and models are available at https://github.com/AntigoneRandy/PTYNet.

Causal inference is a science with multi-disciplinary evolution and applications. On the one hand, it measures effects of treatments in observational data based on experimental designs and rigorous statistical inference to draw causal statements. One of the most influential framework in quantifying causal effects is the potential outcomes framework. On the other hand, causal graphical models utilizes directed edges to represent causalities and encodes conditional independence relationships among variables in the graphs. A series of research has been done both in reading-off conditional independencies from graphs and in re-constructing causal structures. In recent years, the most state-of-art research in causal inference starts unifying the different causal inference frameworks together. This survey aims to provide a review of the past work on causal inference, focusing mainly on potential outcomes framework and causal graphical models. We hope that this survey will help accelerate the understanding of causal inference in different domains.

DeepFake is becoming a real risk to society and brings potential threats to both individual privacy and political security due to the DeepFaked multimedia are realistic and convincing. However, the popular DeepFake passive detection is an ex-post forensics countermeasure and failed in blocking the disinformation spreading in advance. To address this limitation, researchers study the proactive defense techniques by adding adversarial noises into the source data to disrupt the DeepFake manipulation. However, the existing studies on proactive DeepFake defense via injecting adversarial noises are not robust, which could be easily bypassed by employing simple image reconstruction revealed in a recent study MagDR. In this paper, we investigate the vulnerability of the existing forgery techniques and propose a novel \emph{anti-forgery} technique that helps users protect the shared facial images from attackers who are capable of applying the popular forgery techniques. Our proposed method generates perceptual-aware perturbations in an incessant manner which is vastly different from the prior studies by adding adversarial noises that is sparse. Experimental results reveal that our perceptual-aware perturbations are robust to diverse image transformations, especially the competitive evasion technique, MagDR via image reconstruction. Our findings potentially open up a new research direction towards thorough understanding and investigation of perceptual-aware adversarial attack for protecting facial images against DeepFakes in a proactive and robust manner. We open-source our tool to foster future research. Code is available at https://github.com/AbstractTeen/AntiForgery/., Comment: Accepted by IJCAI 2022

For a connected graph $\mathcal{G}=(V,E)$ with $n$ nodes, $m$ edges, and Laplacian matrix $\boldsymbol{{\mathit{L}}}$, a grounded Laplacian matrix $\boldsymbol{{\mathit{L}}}(S)$ of $\mathcal{G}$ is a $(n-k) \times (n-k)$ principal submatrix of $\boldsymbol{{\mathit{L}}}$, obtained from $\boldsymbol{{\mathit{L}}}$ by deleting $k$ rows and columns corresponding to $k$ selected nodes forming a set $S\subseteq V$. The smallest eigenvalue $\lambda(S)$ of $\boldsymbol{{\mathit{L}}}(S)$ plays a pivotal role in various dynamics defined on $\mathcal{G}$. For example, $\lambda(S)$ characterizes the convergence rate of leader-follower consensus, as well as the effectiveness of a pinning scheme for the pinning control problem, with larger $\lambda(S)$ corresponding to smaller convergence time or better effectiveness of a pinning scheme. In this paper, we focus on the problem of optimally selecting a subset $S$ of fixed $k \ll n$ nodes, in order to maximize the smallest eigenvalue $\lambda(S)$ of the grounded Laplacian matrix $\boldsymbol{{\mathit{L}}}(S)$. We show that this optimization problem is NP-hard and that the objective function is non-submodular but monotone. Due to the difficulty to obtain the optimal solution, we first propose a na\"{\i}ve heuristic algorithm selecting one optimal node at each time for $k$ iterations. Then we propose a fast heuristic scalable algorithm to approximately solve this problem, using derivative matrix, matrix perturbations, and Laplacian solvers as tools. Our na\"{\i}ve heuristic algorithm takes $\tilde{O}(knm)$ time, while the fast greedy heuristic has a nearly linear time complexity of $\tilde{O}(km)$. We also conduct numerous experiments on different networks sized up to one million nodes, demonstrating the superiority of our algorithm in terms of efficiency and effectiveness.

Variable selection in ultra-high dimensional linear regression is often preceded by a screening step to significantly reduce the dimension. Here a Bayesian variable screening method (BITS) is developed. BITS can successfully integrate prior knowledge, if any, on effect sizes, and the number of true variables. BITS iteratively includes potential variables with the highest posterior probability accounting for the already selected variables. It is implemented by a fast Cholesky update algorithm and is shown to have the screening consistency property. BITS is built based on a model with Gaussian errors, yet, the screening consistency is proved to hold under more general tail conditions. The notion of posterior screening consistency allows the resulting model to provide a good starting point for further Bayesian variable selection methods. A new screening consistent stopping rule based on posterior probability is developed. Simulation studies and real data examples are used to demonstrate scalability and fine screening performance.