Yang Bai’s research while affiliated with Chengdu University of Information Technology and other places

Distant metastasis is the major cause of death in patients with breast cancer. Epithelial–mesenchymal transition (EMT) contributes to breast cancer metastasis. Regulator of G protein-signaling (RGS) proteins modulates metastasis in various cancers. This study identified a novel role for RGS10 in EMT and metastasis in breast cancer. RGS10 protein levels were significantly lower in breast cancer tissues compared to normal breast tissues, and deficiency in RGS10 protein predicted a worse prognosis in patients with breast cancer. RGS10 protein levels were lower in the highly aggressive cell line MDA-MB-231 than in the poorly aggressive, less invasive cell lines MCF7 and SKBR3. Silencing RGS10 in SKBR3 cells enhanced EMT and caused SKBR3 cell migration and invasion. The ability of RGS10 to suppress EMT and metastasis in breast cancer was dependent on lipocalin-2 and MIR539-5p . These findings identify RGS10 as a tumor suppressor, prognostic biomarker, and potential therapeutic target for breast cancer.

The advent of video-based Large Language Models (LLMs) has significantly enhanced video understanding. However, it has also raised some safety concerns regarding data protection, as videos can be more easily annotated, even without authorization. This paper introduces Video Watermarking, a novel technique to protect videos from unauthorized annotations by such video-based LLMs, especially concerning the video content and description, in response to specific queries. By imperceptibly embedding watermarks into key video frames with multi-modal flow-based losses, our method preserves the viewing experience while preventing misuse by video-based LLMs. Extensive experiments show that Video Watermarking significantly reduces the comprehensibility of videos with various video-based LLMs, demonstrating both stealth and robustness. In essence, our method provides a solution for securing video content, ensuring its integrity and confidentiality in the face of evolving video-based LLMs technologies.

Large vision-language models (VLMs) such as GPT-4 have achieved exceptional performance across various multi-modal tasks. However, the deployment of VLMs necessitates substantial energy consumption and computational resources. Once attackers maliciously induce high energy consumption and latency time (energy-latency cost) during inference of VLMs, it will exhaust computational resources. In this paper, we explore this attack surface about availability of VLMs and aim to induce high energy-latency cost during inference of VLMs. We find that high energy-latency cost during inference of VLMs can be manipulated by maximizing the length of generated sequences. To this end, we propose verbose images, with the goal of crafting an imperceptible perturbation to induce VLMs to generate long sentences during inference. Concretely, we design three loss objectives. First, a loss is proposed to delay the occurrence of end-of-sequence (EOS) token, where EOS token is a signal for VLMs to stop generating further tokens. Moreover, an uncertainty loss and a token diversity loss are proposed to increase the uncertainty over each generated token and the diversity among all tokens of the whole generated sequence, respectively , which can break output dependency at token-level and sequence-level. Furthermore, a temporal weight adjustment algorithm is proposed, which can effectively balance these losses. Extensive experiments demonstrate that our verbose images can increase the length of generated sequences by 7.87× and 8.56× compared to original images on MS-COCO and ImageNet datasets, which presents potential challenges for various applications. Our code is available at https://github.com/KuofengGao/Verbose_Images.

Distant metastasis is the major cause of death in patients with breast cancer. Epithelial–mesenchymal transition (EMT) contributes to breast cancer metastasis. Regulator of G protein-signaling (RGS) proteins modulate metastasis in various cancers. This study identified a novel role for RGS10 in EMT and metastasis in breast cancer. RGS10 protein levels were significantly lower in breast cancer tissues compared to normal breast tissues, and deficiency in RGS10 protein predicted a worse prognosis in patients with breast cancer. RGS10 protein levels were lower in the highly aggressive cell line MDA-MB-231 than in the poorly aggressive, less invasive cell lines MCF7 and SKBR3. Silencing RGS10 in SKBR3 cells enhanced EMT and caused SKBR3 cell migration and invasion. The ability of RGS10 to suppress EMT and metastasis in breast cancer was dependent on lipocalin-2 and miR-539-5p. These findings identify RGS10 as a tumor suppressor, prognostic biomarker, and potential therapeutic target for breast cancer.

Multi-modal Large Language Models (MLLMs) have recently achieved enhanced performance across various vision-language tasks including visual grounding capabilities. However, the adversarial robustness of visual grounding remains unexplored in MLLMs. To fill this gap, we use referring expression comprehension (REC) as an example task in visual grounding and propose three adversarial attack paradigms as follows. Firstly, untargeted adversarial attacks induce MLLMs to generate incorrect bounding boxes for each object. Besides, exclusive targeted adversarial attacks cause all generated outputs to the same target bounding box. In addition, permuted targeted adversarial attacks aim to permute all bounding boxes among different objects within a single image. Extensive experiments demonstrate that the proposed methods can successfully attack visual grounding capabilities of MLLMs. Our methods not only provide a new perspective for designing novel attacks but also serve as a strong baseline for improving the adversarial robustness for visual grounding of MLLMs.

Diffusion models have been widely deployed in various image generation tasks, demonstrating an extraordinary connection between image and text modalities. However, they face challenges of being maliciously exploited to generate harmful or sensitive images by appending a specific suffix to the original prompt. Existing works mainly focus on using single-modal information to conduct attacks, which fails to utilize multi-modal features and results in less than satisfactory performance. Integrating multi-modal priors (MMP), i.e. both text and image features, we propose a targeted attack method named MMP-Attack in this work. Specifically, the goal of MMP-Attack is to add a target object into the image content while simultaneously removing the original object. The MMP-Attack shows a notable advantage over existing works with superior universality and transferability, which can effectively attack commercial text-to-image (T2I) models such as DALL-E 3. To the best of our knowledge, this marks the first successful attempt of transfer-based attack to commercial T2I models. Our code is publicly available at https://github.com/ydc123/MMP-Attack.

Adversarial training has shown promise in building robust models against adversarial examples. A major drawback of adversarial training is the computational overhead introduced by the generation of adversarial examples. To overcome this limitation, adversarial training based on single-step attacks has been explored. Previous work improves the single-step adversarial training from different perspectives, e.g., sample initialization, loss regularization, and training strategy. Almost all of them treat the underlying model as a black box. In this work, we propose to exploit the interior building blocks of the model to improve efficiency. Specifically, we propose to dynamically sample lightweight subnetworks as a surrogate model during training. By doing this, both the forward and backward passes can be accelerated for efficient adversarial training. Besides, we provide theoretical analysis to show the model robustness can be improved by the single-step adversarial training with sampled subnetworks. Furthermore, we propose a novel sampling strategy where the sampling varies from layer to layer and from iteration to iteration. Compared with previous methods, our method not only reduces the training cost but also achieves better model robustness. Evaluations on a series of popular datasets demonstrate the effectiveness of the proposed FB-Better. Our code has been released at https://github.com/jiaxiaojunQAQ/ FP-Better.

Backdoor defenses have been studied to alleviate the threat of deep neural networks (DNNs) being backdoor attacked and thus maliciously altered. Since DNNs usually adopt some external training data from an untrusted third party, a robust backdoor defense strategy during the training stage is of importance. We argue that the core of training-time defense is to select poisoned samples and to handle them properly. In this work, we summarize the training-time defenses from a unified framework as splitting the poisoned dataset into two data pools. Under our framework, we propose an adaptively splitting dataset-based defense (ASD). Concretely, we apply loss-guided split and meta-learning-inspired split to dynamically update two data pools. With the split clean data pool and polluted data pool, ASD successfully defends against backdoor attacks during training. Extensive experiments on multiple benchmark datasets and DNN models against six state-of-the-art backdoor attacks demonstrate the superiority of our ASD. Our code is available at https://github.com/KuofengGao/ASD.

Background: More than half of Chinese patients with hormone receptor positive (HR+) ductal carcinoma in situ (DCIS) are treated with mastectomy, and usually subjected to postoperative endocrine therapy (ET). Given that long-term ET can cause severe adverse effects it is important to determine the beneficial effect and safety of post-mastectomy ET on the disease-free survival (DFS) and adverse events in patients with HR+ DCIS. Methods: To explore beneficial effect and safety of post-mastectomy ET in patients with HR+ DCIS, we performed a multicenter, population-based study. This retrospective study analyzed the DFS and adverse events in 1037 HR+ DCIS Chinese patients with or without post-mastectomy ET from eight breast centers between 2006 and 2016. The median follow-up time period was 86 months. Results: There were 791 DCIS patients receiving ET (ET group). Those patients were followed up for a median of 86 months (range, 60-177 months). There were 23 cases with tumor recurrence or distant metastasis. There were similar 5-year DFS rates and DFS between the ET and non-ET groups, even for those with high-risk factors. Conversely, 37.04% of patients suffered from adverse events after ET, which were significantly higher than those in the non-ET group. Conclusions: ET after mastectomy did not benefit patients with HR+ DCIS for their DFS, rather increased adverse events in those patients. Therefore, ET after mastectomy may not be recommended for patients with HR+ DCIS, even for those with high-risk factors, such as multifocal, microinvasive, and higher T stage. Funding: This study was supported by grants from Outstanding Scientific Fund of Shengjing Hospital (201803) and Outstanding Young Scholars of Liaoning Province (2019-YQ-10).