Sequential recommendation is an extensively explored approach to capturing users’ evolving preferences based on past interactions, aimed at predicting their next likely choice. Despite significant advancements in this domain, including methods based on RNNs and self-attention, challenges like limited supervised signals and noisy data caused by unintentional clicks persist. To address these challenges, some studies have incorporated unsupervised learning by leveraging local item contexts within individual sequences. However, these methods often overlook the intricate associations between items across multiple sequences and are susceptible to noise in item co-occurrence patterns. In this context, we introduce a novel framework, Global Unsupervised Data-Augmentation (UDA4SR), which adopts a graph contrastive learning perspective to generate more robust item embeddings for sequential recommendation. Our approach begins by integrating Generative Adversarial Networks (GANs) for data augmentation, which serves as the first step to enhance the diversity and richness of the training data. Then, we build a Global Item Relationship Graph (GIG) based on all user interaction sequences. Subsequently, we employ graph contrastive learning on the refined graph to enhance item embeddings by capturing complex global associations. To model users’ dynamic and diverse interests more effectively, we enhance the CapsNet module with a novel target-attention mechanism. Extensive experiments show that UDA4SR significantly outperforms state-of-the-art approaches.

In the field of computer vision, multimodal image generation has become a research hotspot, especially the task of integrating text, image, and style. In this study, we propose a multimodal image generation method based on Generative Adversarial Networks (GAN), capable of effectively combining text descriptions, reference images, and style information to generate images that meet multimodal requirements. This method involves the design of a text encoder, an image feature extractor, and a style integration module, ensuring that the generated images maintain high quality in terms of visual content and style consistency. We also introduce multiple loss functions, including adversarial loss, text-image consistency loss, and style matching loss, to optimize the generation process. Experimental results show that our method produces images with high clarity and consistency across multiple public datasets, demonstrating significant performance improvements compared to existing methods. The outcomes of this study provide new insights into multimodal image generation and present broad application prospects.

Multimodal image analysis is a significant research direction in the field of computer vision, playing a crucial role in tasks such as image captioning and visual question answering (VQA). However, existing visual-language fusion methods often struggle to capture the fine-grained interactions between visual and language modalities, leading to suboptimal fusion results. To address this issue, this paper proposes a visual-language fusion model based on the Cross-Attention Transformer, which constructs deep interactive relationships between visual and language modalities through cross-attention mechanisms, thereby achieving effective multimodal feature fusion. The proposed model first utilizes convolutional neural networks (CNN) and pre-trained language models (e.g., BERT) to extract visual and language features separately, and then applies cross-attention modules to capture mutual dependencies in feature sequences, resulting in a unified multimodal representation vector. Experimental results demonstrate that the proposed model significantly outperforms traditional methods in tasks such as image captioning and VQA, validating its superiority in multimodal image analysis. Additionally, visualization analysis and ablation experiments further explore the contribution of the cross-attention mechanism to model performance, while discussing the model's limitations and potential future improvements.

This paper presents the Multimodal Language-Image Fusion Network (MLIF-Net), a new architecture to distinguish AI-generated images from real ones. MLIF-Net combines Vision Transformer (ViT) and Large Language Models (LLMs) to build a multimodal feature fusion network that improves AI-generated content detection accuracy. The model uses a Cross-Attention Mechanism to combine visual and semantic features and a Multiscale Contextual Reasoning Layer to capture both global and local image features. An Adaptive Loss Function improves the consistency and robustness of feature extraction. Experimental results show that MLIF-Net outperforms existing models in accuracy, recall, and Average Precision (AP). This approach can lead to more accurate detection of AI-generated content and may have applications in other generative content tasks.

This paper introduces HijackNet, an adversarial prompt hijacking framework designed to improve hijacking success rates and generalization in multi-lingual tasks. Combining Neural Architecture Search (NAS), Reinforcement Learning (RL), and Multi-Objective Optimization (MOO), HijackNet generates adversarial prompts that adapt to target tasks while bypassing model defense mechanisms. This framework optimizes hijacking success, defense evasion, and task completion through a combination of a prompt generator and adversarial optimizer. We introduce a Defense Robustness Score (DRS) to evaluate the stability of prompts in adversarial settings. HijackNet improves adversarial prompt generation, advancing the security and robustness of large language models in adversarial environments.