arXiv subjects

Ruoxi Chen

Publications and source records attributed to Ruoxi Chen.

30 records · Page 2Linked to original sources

arXiv2024

EditShield: Protecting Unauthorized Image Editing by Instruction-guided Diffusion Models

Text-to-image diffusion models have emerged as an evolutionary for producing creative content in image synthesis. Based on the impressive generation abilities of these models, instruction-guided diffusion models can edit images with simple instructions and input images. While they empower users to obtain their desired edited images with ease, they have raised concerns about unauthorized image manipulation. Prior research has delved into the unauthorized use of personalized diffusion models; however, this problem of instruction-guided diffusion models remains largely unexplored. In this paper, we first propose a protection method EditShield against unauthorized modifications from such models. Specifically, EditShield works by adding imperceptible perturbations that can shift the latent representation used in the diffusion process, tricking models into generating unrealistic images with mismatched subjects. Our extensive experiments demonstrate EditShield's effectiveness among synthetic and real-world datasets. Besides, we found that EditShield performs robustly against various manipulation settings across editing types and synonymous instruction phrases.

cs.CR↗

arXiv2024

MLLM-as-a-Judge: Assessing Multimodal LLM-as-a-Judge with Vision-Language Benchmark

Multimodal Large Language Models (MLLMs) have gained significant attention recently, showing remarkable potential in artificial general intelligence. However, assessing the utility of MLLMs presents considerable challenges, primarily due to the absence of multimodal benchmarks that align with human preferences. Drawing inspiration from the concept of LLM-as-a-Judge within LLMs, this paper introduces a novel benchmark, termed MLLM-as-a-Judge, to assess the ability of MLLMs in assisting judges across diverse modalities, encompassing three distinct tasks: Scoring Evaluation, Pair Comparison, and Batch Ranking. Our study reveals that, while MLLMs demonstrate remarkable human-like discernment in Pair Comparison, there is a significant divergence from human preferences in Scoring Evaluation and Batch Ranking. Furthermore, a closer examination reveals persistent challenges in the judgment capacities of LLMs, including diverse biases, hallucinatory responses, and inconsistencies in judgment, even in advanced models such as GPT-4V. These findings emphasize the pressing need for enhancements and further research efforts to be undertaken before regarding MLLMs as fully reliable evaluators. In light of this, we advocate for additional efforts dedicated to supporting the continuous development within the domain of MLLM functioning as judges. The code and dataset are publicly available at our project homepage: \url{https://mllm-judge.github.io/}.

cs.CL↗

arXiv2024

Convex Combination Consistency between Neighbors for Weakly-supervised Action Localization

Weakly-supervised temporal action localization (WTAL) intends to detect action instances with only weak supervision, e.g., video-level labels. The current~\textit{de facto} pipeline locates action instances by thresholding and grouping continuous high-score regions on temporal class activation sequences. In this route, the capacity of the model to recognize the relationships between adjacent snippets is of vital importance which determines the quality of the action boundaries. However, it is error-prone since the variations between adjacent snippets are typically subtle, and unfortunately this is overlooked in the literature. To tackle the issue, we propose a novel WTAL approach named Convex Combination Consistency between Neighbors (C$^3$BN). C$^3$BN consists of two key ingredients: a micro data augmentation strategy that increases the diversity in-between adjacent snippets by convex combination of adjacent snippets, and a macro-micro consistency regularization that enforces the model to be invariant to the transformations~\textit{w.r.t.} video semantics, snippet predictions, and snippet representations. Consequently, fine-grained patterns in-between adjacent snippets are enforced to be explored, thereby resulting in a more robust action boundary localization. Experimental results demonstrate the effectiveness of C$^3$BN on top of various baselines for WTAL with video-level and point-level supervisions. Code is at https://github.com/Qinying-Liu/C3BN.

cs.CV↗

arXiv2024

Sora: A Review on Background, Technology, Limitations, and Opportunities of Large Vision Models

Sora is a text-to-video generative AI model, released by OpenAI in February 2024. The model is trained to generate videos of realistic or imaginative scenes from text instructions and show potential in simulating the physical world. Based on public technical reports and reverse engineering, this paper presents a comprehensive review of the model's background, related technologies, applications, remaining challenges, and future directions of text-to-video AI models. We first trace Sora's development and investigate the underlying technologies used to build this "world simulator". Then, we describe in detail the applications and potential impact of Sora in multiple industries ranging from film-making and education to marketing. We discuss the main challenges and limitations that need to be addressed to widely deploy Sora, such as ensuring safe and unbiased video generation. Lastly, we discuss the future development of Sora and video generation models in general, and how advancements in the field could enable new ways of human-AI interaction, boosting productivity and creativity of video generation.

cs.CV↗

arXiv2023

AdvCheck: Characterizing Adversarial Examples via Local Gradient Checking

Deep neural networks (DNNs) are vulnerable to adversarial examples, which may lead to catastrophe in security-critical domains. Numerous detection methods are proposed to characterize the feature uniqueness of adversarial examples, or to distinguish DNN's behavior activated by the adversarial examples. Detections based on features cannot handle adversarial examples with large perturbations. Besides, they require a large amount of specific adversarial examples. Another mainstream, model-based detections, which characterize input properties by model behaviors, suffer from heavy computation cost. To address the issues, we introduce the concept of local gradient, and reveal that adversarial examples have a quite larger bound of local gradient than the benign ones. Inspired by the observation, we leverage local gradient for detecting adversarial examples, and propose a general framework AdvCheck. Specifically, by calculating the local gradient from a few benign examples and noise-added misclassified examples to train a detector, adversarial examples and even misclassified natural inputs can be precisely distinguished from benign ones. Through extensive experiments, we have validated the AdvCheck's superior performance to the state-of-the-art (SOTA) baselines, with detection rate ($\sim \times 1.2$) on general adversarial attacks and ($\sim \times 1.4$) on misclassified natural inputs on average, with average 1/500 time cost. We also provide interpretable results for successful detection.

cs.CR↗

arXiv2023

Free-electron Brewster radiation

Free-electron radiation offers an enticing route to create light emission at arbitrary spectral regime. However, this type of light emission is generally weak, which is intrinsically limited by the weak particle-matter interaction and unavoidably impedes the development of many promising applications, such as the miniaturization of free-electron radiation sources and high-energy particle detectors. Here we reveal a mechanism to enhance the particle-matter interaction by exploiting the pseudo-Brewster effect of gain materials - presenting an enhancement of at least four orders of magnitude for the light emission. This mechanism is enabled by the emergence of an unprecedented phase diagram that maps all phenomena of free-electron radiation into three distinct phases in a gain-thickness parameter space, namely the conventional, intermediate, and Brewster phases, when an electron penetrates a dielectric slab with a modest gain and a finite thickness. Essentially, our revealed mechanism corresponds to the free-electron radiation in the Brewster phase, which also uniquely features ultrahigh directionality, always at the Brewster angle, regardless of the electron velocity. Counterintuitively, we find that the intensity of this free-electron Brewster radiation is insensitive to the Fabry-Perot resonance condition and thus the variation of slab thickness, and moreover, a weaker gain could lead to a stronger enhancement for the light emission. The scheme of free-electron Brewster radiation, especially along with its compatibility with low-energy electrons, may enable the development of high-directionality high-intensity light sources at any frequency.

physics.optics↗

arXiv2023

Recent advances of transition radiation: fundamentals and applications

Transition radiation is a fundamental process of light emission and occurs whenever a charged particle moves across an inhomogeneous region. One feature of transition radiation is that it can create light emission at arbitrary frequency under any particle velocity. Therefore, transition radiation is of significant importance to both fundamental science and practical applications. In this paper, we provide a brief historical review of transition radiation and its recent development. Moreover, we pay special attention to four typical applications of transition radiation, namely the detection of high-energy particles, coherent radiation sources, beam diagnosis, and excitation of surface waves. Finally, we give an outlook for the research tendency of transition radiation, especially its flexible manipulation by exploiting artificially-engineered materials and nanostructures, such as gain materials, metamaterials, spatial-temporal materials, meta-boundaries, and layered structures with a periodic or non-periodic stacking.

physics.optics↗

arXiv2022

Low-velocity-favored transition radiation

When a charged particle penetrates through an optical interface, photon emissions emerge - a phenomenon known as transition radiation. Being paramount to fundamental physics, transition radiation has enabled many applications from high-energy particle identification to novel light sources. A rule of thumb in transition radiation is that the radiation intensity generally decreases with the particle velocity v; as a result, low-energy particles are not favored in practice. Here we find that there exist situations where transition radiation from particles with extremely low velocities (e.g. v/c<0.001) exhibits comparable intensity as that from high-energy particles (e.g. v/c=0.999), where c is light speed in free space. The comparable radiation intensity implies an extremely high photon extraction efficiency from low-energy particles, up to eight orders of magnitude larger than that from high-energy particles. This exotic phenomenon of low-velocity-favored transition radiation originates from the excitation of Ferrell-Berreman modes in epsilon-near-zero materials. Our findings may provide a promising route towards the design of integrated light sources based on low-energy electrons and specialized detectors for beyond-standard-model particles.

physics.optics↗

arXiv2022

Excitement Surfeited Turns to Errors: Deep Learning Testing Framework Based on Excitable Neurons

Despite impressive capabilities and outstanding performance, deep neural networks (DNNs) have captured increasing public concern about their security problems, due to their frequently occurred erroneous behaviors. Therefore, it is necessary to conduct a systematical testing for DNNs before they are deployed to real-world applications. Existing testing methods have provided fine-grained metrics based on neuron coverage and proposed various approaches to improve such metrics. However, it has been gradually realized that a higher neuron coverage does \textit{not} necessarily represent better capabilities in identifying defects that lead to errors. Besides, coverage-guided methods cannot hunt errors due to faulty training procedure. So the robustness improvement of DNNs via retraining by these testing examples are unsatisfactory. To address this challenge, we introduce the concept of excitable neurons based on Shapley value and design a novel white-box testing framework for DNNs, namely DeepSensor. It is motivated by our observation that neurons with larger responsibility towards model loss changes due to small perturbations are more likely related to incorrect corner cases due to potential defects. By maximizing the number of excitable neurons concerning various wrong behaviors of models, DeepSensor can generate testing examples that effectively trigger more errors due to adversarial inputs, polluted data and incomplete training. Extensive experiments implemented on both image classification models and speaker recognition models have demonstrated the superiority of DeepSensor.

cs.LG↗

arXiv2022

A perspective on meta-boundaries

The judicious design of electromagnetic boundary provides a crucial route to control light-matter interactions, and it is thus fundamental to basic science and practical applications. General design approaches rely on the manipulation of bulk properties of superstrate or substrate and on the modification of boundary geometries. Due to the recent advent of metasurfaces and low-dimensional materials, the boundary can be flexibly featured with a surface conductivity, which can be rather complex but provide an extra degree of freedom to regulate the propagation of light. In this perspective, we denote the boundary with a non-zero surface conductivity as the meta-boundary. The meta-boundaries are categorized into four types, namely isotropic, anisotropic, biisotropic and bianisotropic meta-boundaries, according to the electromagnetic boundary conditions. Accordingly, the latest development for these four kinds of meta-boundaries are reviewed. Finally, an outlook on the research tendency of meta-boundaries is provided, particularly on the manipulation of light-matter interactions by simultaneously exploiting meta-boundaries and metamaterials.

physics.optics↗

arXiv2022

Is Multi-Modal Necessarily Better? Robustness Evaluation of Multi-modal Fake News Detection

The proliferation of fake news and its serious negative social influence push fake news detection methods to become necessary tools for web managers. Meanwhile, the multi-media nature of social media makes multi-modal fake news detection popular for its ability to capture more modal features than uni-modal detection methods. However, current literature on multi-modal detection is more likely to pursue the detection accuracy but ignore the robustness of the detector. To address this problem, we propose a comprehensive robustness evaluation of multi-modal fake news detectors. In this work, we simulate the attack methods of malicious users and developers, i.e., posting fake news and injecting backdoors. Specifically, we evaluate multi-modal detectors with five adversarial and two backdoor attack methods. Experiment results imply that: (1) The detection performance of the state-of-the-art detectors degrades significantly under adversarial attacks, even worse than general detectors; (2) Most multi-modal detectors are more vulnerable when subjected to attacks on visual modality than textual modality; (3) Popular events' images will cause significant degradation to the detectors when they are subjected to backdoor attacks; (4) The performance of these detectors under multi-modal attacks is worse than under uni-modal attacks; (5) Defensive methods will improve the robustness of the multi-modal detectors.

cs.AI↗

arXiv2021

Salient Feature Extractor for Adversarial Defense on Deep Neural Networks

Recent years have witnessed unprecedented success achieved by deep learning models in the field of computer vision. However, their vulnerability towards carefully crafted adversarial examples has also attracted the increasing attention of researchers. Motivated by the observation that adversarial examples are due to the non-robust feature learned from the original dataset by models, we propose the concepts of salient feature(SF) and trivial feature(TF). The former represents the class-related feature, while the latter is usually adopted to mislead the model. We extract these two features with coupled generative adversarial network model and put forward a novel detection and defense method named salient feature extractor (SFE) to defend against adversarial attacks. Concretely, detection is realized by separating and comparing the difference between SF and TF of the input. At the same time, correct labels are obtained by re-identifying SF to reach the purpose of defense. Extensive experiments are carried out on MNIST, CIFAR-10, and ImageNet datasets where SFE shows state-of-the-art results in effectiveness and efficiency compared with baselines. Furthermore, we provide an interpretable understanding of the defense and detection process.

cs.CV↗