Xiaodan Zhang

Yiming Huang, Aozhe Jia, Xiaodan Zhang et al.

Attention-based transformer models have achieved remarkable progress in multi-modal tasks, such as visual question answering. The explainability of attention-based methods has recently attracted wide interest as it can explain the inner changes of attention tokens by accumulating relevancy across attention layers. Current methods simply update relevancy by equally accumulating the token relevancy before and after the attention processes. However, the importance of token values is usually different during relevance accumulation. In this paper, we propose a weighted relevancy strategy, which takes the importance of token values into consideration, to reduce distortion when equally accumulating relevance. To evaluate our method, we propose a unified CLIP-based two-stage model, named CLIPmapper, to process Vision-and-Language tasks through CLIP encoder and a following mapper. CLIPmapper consists of self-attention, cross-attention, single-modality, and cross-modality attention, thus it is more suitable for evaluating our generic explainability method. Extensive perturbation tests on visual question answering and image captioning validate that our explainability method outperforms existing methods.

Bingkang Shi, Xiaodan Zhang, Dehan Kong et al.

The social biases and unwelcome stereotypes revealed by pretrained language models are becoming obstacles to their application. Compared to numerous debiasing methods targeting word level, there has been relatively less attention on biases present at phrase level, limiting the performance of debiasing in discipline domains. In this paper, we propose an automatic multi-token debiasing pipeline called \textbf{General Phrase Debiaser}, which is capable of mitigating phrase-level biases in masked language models. Specifically, our method consists of a \textit{phrase filter stage} that generates stereotypical phrases from Wikipedia pages as well as a \textit{model debias stage} that can debias models at the multi-token level to tackle bias challenges on phrases. The latter searches for prompts that trigger model's bias, and then uses them for debiasing. State-of-the-art results on standard datasets and metrics show that our approach can significantly reduce gender biases on both career and multiple disciplines, across models with varying parameter sizes.

Wenchao Zhang, Jiahe Tian, Runze He et al.

Recent text-to-image (T2I) generation models have advanced significantly, enabling the creation of high-fidelity images from textual prompts. However, existing evaluation benchmarks primarily focus on the explicit alignment between generated images and prompts, neglecting the alignment with real-world knowledge beyond prompts. To address this gap, we introduce Align Beyond Prompts (ABP), a comprehensive benchmark designed to measure the alignment of generated images with real-world knowledge that extends beyond the explicit user prompts. ABP comprises over 2,000 meticulously crafted prompts, covering real-world knowledge across six distinct scenarios. We further introduce ABPScore, a metric that utilizes existing Multimodal Large Language Models (MLLMs) to assess the alignment between generated images and world knowledge beyond prompts, which demonstrates strong correlations with human judgments. Through a comprehensive evaluation of 8 popular T2I models using ABP, we find that even state-of-the-art models, such as GPT-4o, face limitations in integrating simple real-world knowledge into generated images. To mitigate this issue, we introduce a training-free strategy within ABP, named Inference-Time Knowledge Injection (ITKI). By applying this strategy to optimize 200 challenging samples, we achieved an improvement of approximately 43% in ABPScore. The dataset and code are available in https://github.com/smile365317/ABP.

Shang Shang, Xinqiang Zhao, Zhongjiang Yao et al.

To demonstrate and address the underlying maliciousness, we propose a theoretical hypothesis and analytical approach, and introduce a new black-box jailbreak attack methodology named IntentObfuscator, exploiting this identified flaw by obfuscating the true intentions behind user prompts.This approach compels LLMs to inadvertently generate restricted content, bypassing their built-in content security measures. We detail two implementations under this framework: "Obscure Intention" and "Create Ambiguity", which manipulate query complexity and ambiguity to evade malicious intent detection effectively. We empirically validate the effectiveness of the IntentObfuscator method across several models, including ChatGPT-3.5, ChatGPT-4, Qwen and Baichuan, achieving an average jailbreak success rate of 69.21\%. Notably, our tests on ChatGPT-3.5, which claims 100 million weekly active users, achieved a remarkable success rate of 83.65\%. We also extend our validation to diverse types of sensitive content like graphic violence, racism, sexism, political sensitivity, cybersecurity threats, and criminal skills, further proving the substantial impact of our findings on enhancing 'Red Team' strategies against LLM content security frameworks.

Xiaowei Yuan, Jingyuan Hu, Xiaodan Zhang et al.

Emojis are widely used in online social networks to express emotions, attitudes, and opinions. As emotional-oriented characters, emojis can be modeled as important features of emotions towards the recipient or subject for sentiment analysis. However, existing methods mainly take emojis as heuristic information that fails to resolve the problem of ambiguity noise. Recent researches have utilized emojis as an independent input to classify text sentiment but they ignore the emotional impact of the interaction between text and emojis. It results that the emotional semantics of emojis cannot be fully explored. In this paper, we propose an emoji-based co-attention network that learns the mutual emotional semantics between text and emojis on microblogs. Our model adopts the co-attention mechanism based on bidirectional long short-term memory incorporating the text and emojis, and integrates a squeeze-and-excitation block in a convolutional neural network classifier to increase its sensitivity to emotional semantic features. Experimental results show that the proposed method can significantly outperform several baselines for sentiment analysis on short texts of social media.

Lingwei Wei, Dou Hu, Wei Zhou et al.

Document-level Sentiment Analysis (DSA) is more challenging due to vague semantic links and complicate sentiment information. Recent works have been devoted to leveraging text summarization and have achieved promising results. However, these summarization-based methods did not take full advantage of the summary including ignoring the inherent interactions between the summary and document. As a result, they limited the representation to express major points in the document, which is highly indicative of the key sentiment. In this paper, we study how to effectively generate a discriminative representation with explicit subject patterns and sentiment contexts for DSA. A Hierarchical Interaction Networks (HIN) is proposed to explore bidirectional interactions between the summary and document at multiple granularities and learn subject-oriented document representations for sentiment classification. Furthermore, we design a Sentiment-based Rethinking mechanism (SR) by refining the HIN with sentiment label information to learn a more sentiment-aware document representation. We extensively evaluate our proposed models on three public datasets. The experimental results consistently demonstrate the effectiveness of our proposed models and show that HIN-SR outperforms various state-of-the-art methods.

Chunyuan Yuan, Jiacheng Li, Wei Zhou et al.

Information diffusion prediction is a fundamental task for understanding the information propagation process. It has wide applications in such as misinformation spreading prediction and malicious account detection. Previous works either concentrate on utilizing the context of a single diffusion sequence or using the social network among users for information diffusion prediction. However, the diffusion paths of different messages naturally constitute a dynamic diffusion graph. For one thing, previous works cannot jointly utilize both the social network and diffusion graph for prediction, which is insufficient to model the complexity of the diffusion process and results in unsatisfactory prediction performance. For another, they cannot learn users' dynamic preferences. Intuitively, users' preferences are changing as time goes on and users' personal preference determines whether the user will repost the information. Thus, it is beneficial to consider users' dynamic preferences in information diffusion prediction. In this paper, we propose a novel dynamic heterogeneous graph convolutional network (DyHGCN) to jointly learn the structural characteristics of the social graph and dynamic diffusion graph. Then, we encode the temporal information into the heterogeneous graph to learn the users' dynamic preferences. Finally, we apply multi-head attention to capture the context-dependency of the current diffusion path to facilitate the information diffusion prediction task. Experimental results show that DyHGCN significantly outperforms the state-of-the-art models on three public datasets, which shows the effectiveness of the proposed model.

Xiaodan Zhang, Xinbo Gao, Wen Lu et al.

Learning fine-grained details is a key issue in image aesthetic assessment. Most of the previous methods extract the fine-grained details via random cropping strategy, which may undermine the integrity of semantic information. Extensive studies show that humans perceive fine-grained details with a mixture of foveal vision and peripheral vision. Fovea has the highest possible visual acuity and is responsible for seeing the details. The peripheral vision is used for perceiving the broad spatial scene and selecting the attended regions for the fovea. Inspired by these observations, we propose a Gated Peripheral-Foveal Convolutional Neural Network (GPF-CNN). It is a dedicated double-subnet neural network, i.e. a peripheral subnet and a foveal subnet. The former aims to mimic the functions of peripheral vision to encode the holistic information and provide the attended regions. The latter aims to extract fine-grained features on these key regions. Considering that the peripheral vision and foveal vision play different roles in processing different visual stimuli, we further employ a gated information fusion (GIF) network to weight their contributions. The weights are determined through the fully connected layers followed by a sigmoid function. We conduct comprehensive experiments on the standard AVA and Photo.net datasets for unified aesthetic prediction tasks: (i) aesthetic quality classification; (ii) aesthetic score regression; and (iii) aesthetic score distribution prediction. The experimental results demonstrate the effectiveness of the proposed method.