Yongdong Zhang

In the era of Large Language Models (LLMs), the Internet is flooded with LLM-generated content, including both aligned LLM-generated content that contains equivalent facts to human-written content and distorted LLM-generated content with factual discrepancies. However, there is a significant gap in understanding and addressing the impact of such content on Retrieval-Augmented Generation (RAG) systems, which rely on accurate retrieved knowledge to generate response. This article comprehensively evaluate the impact of aligned and distorted LLM-generated content on the retrieval and generation stages of RAG systems. Notably, we reveal that distorted LLM-generated content is not only more likely to be retrieved but also significantly degrades generation quality, whereas aligned content can, in some cases, enhance it. To mitigate this, we propose a factual consistency-aware adaptive filtering approach to selectively filter out distorted LLM-generated content from retrieved documents. Experimental results demonstrate that our method significantly improves generative performance in scenarios that mix with LLM-generated content and is broadly applicable across various RAG systems.

Face swapping efforts strive to achieve high-fidelity and well-controlled generation effects. Owing to the remarkable generative capabilities, diffusion models deliver promising high-fidelity solutions. However, their intrinsic stochastic properties complicate the accurate modeling of facial representations, introducing new challenges for identity and attribute consistency of the generated faces. In this paper, we introduce a novel diffusion-based face-swapping framework, named SwapController, which achieves high-fidelity generation via careful facial identity and attribute modeling. Specifically, our facial modeling mainly involves facial structure and facial texture. For structure modeling, 3D facial priors are leveraged to provide explicit structure supervision, enabling accurate head structure control. On this basis, two novel components are proposed to deeply mine facial textural representations from identity and attribute aspects. To enhance identity control, multi-grained source identity embeddings are obtained from various functional encoders to convey critical global identity and fine-grained identity details. To improve attribute modeling, identity-shifted attribute embeddings are derived by applying identity modulation to the most salient textural attribute features of the target face. Moreover, in line with the diffusion denoising characteristics, a timestep-aware identity optimization objective is introduced to optimize identity consistency guidelines and overall fidelity. Extensive experiments demonstrate the effectiveness of our SwapController in generating identity-consistent portrait images while faithfully preserving target attributes, which obtains a 98.32 ID Retrieval, exceeding the SOTA DiffSFSR by 7.32 $\uparrow$. Codes will be available at https://github.com/tyrink/SwapController.

Proactive defense planning in Industrial Control Systems (ICS) is critically constrained by two intertwined types of uncertainty: epistemic uncertainty, arising from the defender’s limited observability of the system state and incomplete knowledge of attacker strategies, and aleatoric uncertainty, stemming from the stochastic nature of state transitions and the propagation of disturbances through inter-device dependencies. These factors significantly complicate the quantitative assessment of defense strategies before deployment. To address this challenge, this study proposes a Dynamic Bayesian Network (DBN)-based framework that explicitly models four sources of uncertainty. Within this framework, the expectation of the effectiveness differential is coupled with its information entropy to jointly quantify expected performance and prediction uncertainty. A casestudy on a typical substation automation system, complemented by systematic ablation experiments, demonstrates that the framework can effectively distinguish the relative effectiveness of defense strategies. The framework maintains robust assessment results under up to 15% noise in Conditional Probability Tables (CPTs). The ablation experiments further quantify the individual contributions of observability, dependency propagation, and attacker strategy to prediction uncertainty, and reveal a non-trivial coupling between epistemic and aleatoric uncertainty. This research provides theoretical and methodological support for resilience-oriented cyber defense planning in ICS.

Machine learning has been successfully applied to accelerate Mixed-Integer Linear Programming (MILP) solvers. However, the learning-based solvers often suffer from severe performance degradation on unseen MILP instances due to the limited number of training instances and diversity of training distributions. To tackle this problem, we propose a novel data augmentation approach, called Adversarial Instance Augmentation, to promote the data diversity for learning-based branching modules in branch-and-bound (B&B) Solvers (AdaSolver). To generate diverse augmented instances, AdaSolver uses an augmentation policy to augment the structures of the bipartite graphs representing MILP instances. The major technical contribution is that AdaSolver formulates the non-differentiable learning problem of the augmentation policy as a contextual bandit problem, enabling efficient gradient-based adversarial training for the learning-based solver and augmentation policy. To the best of our knowledge, AdaSolver is the first general and effective framework for understanding and improving the generalization of both imitation-learning-based and reinforcement-learning-based B&B solvers. Experiments demonstrate that AdaSolver leads to a remarkable roughly 35% improvement in solving time over the B&B solver across various distributions. Moreover, AdaSolver significantly improves the sample efficiency by roughly 40% reduction in solving time compared to the competitive GNN branching baseline with only 1% of the training instances.

With the rapid advancement of diffusion models, AI-generated content (AIGC), particularly in image and video generation, has achieved remarkable progress while simultaneously raising significant concerns about content safety within the community. Compared to image generation, the risks of generated videos remain largely overlooked and more challenging, as harmful content may be concealed at any temporal location or accumulate over time. Therefore, this paper proposes a novel Dual-Safety Approach (DSA) to remove temporal risks in generated videos while preserving their inherent quality at test time. The DSA solves the temporal risks from two aspects: the semantics in the text prompts and the temporal dynamics in the diffusion generation process. Specifically, we introduce a Semantic Adaptive Refinement module, which constructs a prompt-specific harmful subspace for targeted risk mitigation while preserving core semantics. Rather than treating video safety at a coarse, whole-video level, we introduce Temporal Dynamic Adaptive Optimization and Temporal Safety Guidance to dynamically constrain unsafe content over time while preserving temporal consistency. Moreover, to evaluate the effectiveness of our approach, we construct a novel Temporal Safety Benchmark for both Text-to-Video and Image-to-Video models. Extensive experiments demonstrate that our DSA can significantly reduce the risks of videos generated by open-source models without degrading video quality and even surpassing commercial models. Additional evaluations on multiple benchmarks, including adversarial attack datasets, further confirm that DSA surpasses existing methods in both effectiveness and robustness. Code and data will be publicly available.

Retrieval-Augmented Generation (RAG) is an effective method to enhance the capabilities of large language models (LLMs). Existing methods typically optimize the retriever or the generator in a RAG system by directly using the top-$k$ retrieved documents. However, two key issues inherent in the training data constrain the effectiveness of this training paradigm: (1) across different queries, the top-$k$ retrieved documents vary greatly in content quality, with some providing valuable knowledge while others lack critical information or are even misleading, and training on such data in a purely random manner may impair the generator's ability to extract key information; (2) for a given query, the limited set of $k$ documents often exhibits low discriminability, and training solely on them makes it difficult for the retriever to learn how to distinguish between relevant and irrelevant documents. To address these issues, we introduce DACL-RAG, a multi-stage RAG training framework that combines a multi-level Data Augmentation strategy with a multi-stage Curriculum Learning paradigm. The data augmentation strategy constructs comprehensive and diverse training sets with controllable difficulty levels through sample evolution, while the curriculum learning paradigm organizes them into progressive stages for training, ensuring stable and consistent improvements, thereby optimizing the overall performance and generalization of the RAG system more effectively. Our DACL-RAG framework demonstrates consistent effectiveness across four open-domain QA datasets, achieving performance gains of 2% to 4% over multiple advanced methods.