YYYing Yang

Ying Yang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Ying Yang has a total of 21 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Semantic and Precise Trigger Inversion: Detecting Backdoored Language Models

    by Chunlong Xie, Jialing He, Ying Yang, Shangwei Guo, Tianwei Zhang, Tao Xiang

    Backdoor attacks pose a serious security threat to Natural Language Processing (NLP) models, allowing adversaries to manipulate model outputs through hidden triggers. Although backdoor detection methods have been developed to address this issue, existing approaches based on trigger inversion are effective only for simple, visible triggers. These methods struggle to handle semantically enhanced, invisible triggers and often fail to provide accurate backdoor determinations due to reliance on unreliable heuristics, making it difficult to reliably distinguish backdoored models from benign ones. This presents a critical gap in current detection techniques. To address these challenges, we propose a novel trigger inversion SemInv that consists of two key contributions: consistent semantics inversion and identifiable condition inspection. Consistent semantics inversion introduces a new regularization technique into the trigger optimization process, enabling more effective inversion of semantically constrained triggers. Identifiable condition inspection assesses the attack performance margin across different identifiable conditions, providing robust evidence for distinguishing backdoored models from benign ones. We evaluate SemInv using the TrojAI round 6–8 datasets and demonstrate that it significantly outperforms state-of-the-art approaches in both backdoor detection accuracy and trigger inversion performance. Our method also proves effective against models with stealthy triggers, advancing the field of NLP security by offering a more comprehensive solution for identifying backdoor attacks. The code repository is in https://github.com/Bluedask/SemInv.


  • Stealthiness Assessment of Adversarial Perturbation: From a Visual Perspective

    by Hangcheng Liu, Yuan Zhou, Ying Yang, Qingchuan Zhao, Tianwei Zhang, Tao Xiang

    Assessing the stealthiness of adversarial perturbations is challenging due to the lack of appropriate evaluation metrics. Existing evaluation metrics, e.g., Lp norms or Image Quality Assessment (IQA), fall short of assessing the pixel-level stealthiness of subtle adversarial perturbations since these metrics are primarily designed for traditional distortions. To bridge this gap, we present the first comprehensive study on the subjective and objective assessment of the stealthiness of adversarial perturbations from a visual perspective at a pixel level. Specifically, we propose new subjective assessment criteria for human observers to score adversarial stealthiness in a fine-grained manner. Then, we create a large-scale adversarial example dataset comprising 10586 pairs of clean and adversarial samples encompassing twelve state-of-the-art adversarial attacks. To obtain the subjective scores according to the proposed criterion, we recruit 60 human observers, and each adversarial example is evaluated by at least 15 observers. The mean opinion score of each adversarial example is utilized for labeling. Finally, we develop a three-stage objective scoring model that mimics human scoring habits to predict adversarial perturbation’s stealthiness. Experimental results demonstrate that our objective model exhibits superior consistency with the human visual system, surpassing commonly employed metrics like PSNR and SSIM.


  • Self-supervised multi-scale neural network for blind deblurring

    by Meina Zhang, Ying Yang, Guoxi Ni, Tingting Wu, Tieyong Zeng

    Blurry kernel estimation is a critical yet challenging task for blind deblurring. Most existing works devote to designing end-to-end networks that require a large amount of hard-to-obtain training data. In addition, these methods often ignore the intrinsic effects of blur kernel for blind deblurring. In this work, we present a unified latent image deblur and kernel estimation method based on MAP framework. By revisiting the coarse-to-fine strategy, we introduce a self-supervised multi-scale deblur network(MD-Net), where the multi-scale structure significantly reduce the kernel deviation caused by local area minimization. Specifically, our network commences with random inputs and outputs multi-scale reconstructed images and kernels. By progressively capturing the high-level configuration and low-level details from matching multi-resolution loss functions, the proposed MD-Net enable to capture multi-level image priors. Meanwhile, at each coarse level, we use Feature Extraction(FE) layers to further extract and emphasize features from reconstructed images. Compared with state-of-the-art blind deblurring methods, extensive experiments demonstrate that the proposed approach significantly improves the restoration performance in both quantitative and qualitative evaluations.


  • Double Transformer Super-Resolution for Breast Cancer ADC Images

    by Ying Yang, Tao Xiang, Xiao Lv, Lihua Li, Lok Ming Lui, Tieyong Zeng

    Diffusion-weighted imaging (DWI) has been extensively explored in guiding the clinic management of patients with breast cancer. However, due to the limited resolution, accurately characterizing tumors using DWI and the corresponding apparent diffusion coefficient (ADC) is still a challenging problem. In this paper, we aim to address the issue of super-resolution (SR) of ADC images and evaluate the clinical utility of SR-ADC images through radiomics analysis. To this end, we propose a novel double transformer-based network (DTformer) to enhance the resolution of ADC images. More specifically, we propose a symmetric U-shaped encoder-decoder network with two different types of transformer blocks, named as UTNet, to extract deep features for super-resolution. The basic backbone of UTNet is composed of a locally-enhanced Swin transformer block (LeSwin-T) and a convolutional transformer block (Conv-T), which are responsible for capturing long-range dependencies and local spatial information, respectively. Additionally, we introduce a residual upsampling network (RUpNet) to expand image resolution by leveraging initial residual information from the original low-resolution (LR) images. Extensive experiments show that DTformer achieves superior SR performance. Moreover, radiomics analysis reveals that improving the resolution of ADC images is beneficial for tumor characteristic prediction, such as histological grade and human epidermal growth factor receptor 2 (HER2) status.


  • The Illusion of Visual Security: Reconstructing Perceptually Encrypted Images

    by Ying Yang, Tao Xiang, Xiao Lv, Shangwei Guo, Tieyong Zeng

    Perceptual image encryption degrades image quality by selectively encrypting some key information of the plain images. The encrypted images are partially perceptible according to the security or quality requirements. Although several types of attacks have tried to infer privacy information from the encrypted images, they can only either extract statistical information or enhance image sketch. In this paper, we take one step further and fully recover the plain images from perceptually encrypted counterparts by designing a non-local attack network (NL-ANet). NL-ANet is composed of densely cascaded multiscale non-local modules (MSNL) and a hierarchical attention fusion module (HAFM). In particular, to better reconstruct encryption distortion, we introduce MSNL to capture powerful hierarchical features from different scales, and propose HAFM to adaptively aggregate and enhance informative hierarchical features for reconstruction. We also propose a new instantiation of the multi-head non-local block with channel attention (MHCA) to explore the long-range dependencies of global contextual information. Extensive experiments show that NL-ANet is encryption-agnostic and superior on different perceptual encryption schemes under different encryption strengths. NL-ANet also achieves better performance than state-of-the-art image restoration methods.


  • Open Access

    Retinex Image Enhancement Based on Sequential Decomposition With a Plug-and-Play Framework

    by Tingting Wu, Wenna Wu, Ying Yang, Feng-Lei Fan, Tieyong Zeng

    The Retinex model is one of the most representative and effective methods for low-light image enhancement. However, the Retinex model does not explicitly tackle the noise problem and shows unsatisfactory enhancing results. In recent years, due to the excellent performance, deep learning models have been widely used in low-light image enhancement. However, these methods have two limitations. First, the desirable performance can only be achieved by deep learning when a large number of labeled data are available. However, it is not easy to curate massive low-/normal-light paired data. Second, deep learning is notoriously a black-box model. It is difficult to explain their inner working mechanism and understand their behaviors. In this article, using a sequential Retinex decomposition strategy, we design a plug-and-play framework based on the Retinex theory for simultaneous image enhancement and noise removal. Meanwhile, we develop a convolutional neural network-based (CNN-based) denoiser into our proposed plug-and-play framework to generate a reflectance component. The final image is enhanced by integrating the illumination and reflectance with gamma correction. The proposed plug-and-play framework can facilitate both post hoc and ad hoc interpretability. Extensive experiments on different datasets demonstrate that our framework outcompetes the state-of-the-art methods in both image enhancement and denoising.


  • Blind Dehazed Image Quality Assessment: A Deep CNN-Based Approach

    by Xiao Lv, Tao Xiang, Ying Yang, Hantao Liu

    Research on image dehazing has made the need for a suitable dehazed image quality assessment (DIQA) method even more urgent. The performance of existing DIQA methods heavily relies on handcrafted haze-related features. Since hazy images with uneven haze density distributions will result in uneven quality distributions after dehazing, the manually extracted feature expression is neither accurate nor robust. In this paper, we design a deep CNN-based DIQA method without a handcrafted feature requirement. Specifically, we propose a blind dehazed image quality assessment model (BDQM), which consists of three components: image preprocessing, a haze-related feature extraction network (HFNet), and an improved regression network (IRNet). In HFNet, we design a perceptual information enhancement (PIE) module to learn powerful feature representations and enhance network capability according to channel attention, multiscale convolution and residual concatenation. IRNet aims to aggregate all patch information for the quality prediction of the whole image, where the effect of inhomogeneous distortion from the dehazing procedure is attenuated via a specifically designed patch attention (PA) mechanism. Experimental results on benchmark datasets demonstrate the effectiveness and superiority of the proposed network architecture over state-of-the-art methods.


  • Open Access

    Retinex Image Enhancement Based on Sequential Decomposition With a Plug-and-Play Framework

    by Tingting Wu, Wenna Wu, Ying Yang, Feng-Lei Fan, Tieyong Zeng

    The Retinex model is one of the most representative and effective methods for low-light image enhancement. However, the Retinex model does not explicitly tackle the noise problem, and shows unsatisfactory enhancing results. In recent years, due to the excellent performance, deep learning models have been widely used in low-light image enhancement. However, these methods have two limitations: i) The desirable performance can only be achieved by deep learning when a large number of labeled data are available. However, it is not easy to curate massive low/normal-light paired data; ii) Deep learning is notoriously a black-box model [1]. It is difficult to explain their inner-working mechanism and understand their behaviors. In this paper, using a sequential Retinex decomposition strategy, we design a plug-and-play framework based on the Retinex theory for simultaneously image enhancement and noise removal. Meanwhile, we develop a convolutional neural network-based (CNN-based) denoiser into our proposed plug-and-play framework to generate a reflectance component. The final enhanced image is produced by integrating the illumination and reflectance with gamma correction. The proposed plug-and-play framework can facilitate both post hoc and ad hoc interpretability. Extensive experiments on different datasets demonstrate that our framework outcompetes the state-of-the-art methods in both image enhancement and denoising.