Optimization modeling and solving are fundamental to the application of Operations Research (OR) in real-world decision making, yet the process of translating natural language problem descriptions into formal models and solver code remains highly expertise intensive. While recent advances in large language models (LLMs) have opened new opportunities for automation, the generalization ability and data efficiency of existing LLM-based methods are still limited, asmost require vast amounts of annotated or synthetic data, resulting in high costs and scalability barriers. In this work, we present OR-R1, a data-efficient training framework for automated optimization modeling and solving. OR-R1 first employs supervised fine-tuning (SFT) to help the model acquire the essential reasoning patterns for problem formulation and code generation from limited labeled data. In addition, it improves the capability and consistency through Test-Time Group Relative Policy Optimization (TGRPO). This two-stage design enables OR-R1 to leverage both scarce labeled and abundant unlabeled data for effective learning. Experiments show that OR-R1 achieves state-of-the-art performance with an average solving accuracy of 67.7%, using only 1/10 the synthetic data required by prior methods such as ORLM, exceeding ORLM’s solving accuracy by up to 4.2%. Remarkably, OR-R1 outperforms ORLM by over 2.4% with just 100 synthetic samples. Furthermore, TGRPO contributes an additional 3.1%–6.4% improvement in accuracy, significantly narrowing the gap between single-attempt (Pass@1) and multi-attempt (Pass@8) performance from 13% to 7%. Extensive evaluations across diverse real-world benchmarks demonstrate that OR-R1 provides a robust, scalable, and cost-effective solution for automated OR optimization problem modeling and solving, lowering the expertise and data barriers for industrial OR applications.

Agents based on Large Language Models (LLMs) have demonstrated strong capabilities across a wide range of tasks. However, deploying LLM-based agents in high-stakes domains comes with significant safety and ethical risks. Unethical behavior by these agents can directly result in serious real-world consequences, including physical harm and financial loss. To efficiently steer the ethical behavior of agents, we frame agent behavior steering as a model editing task, which we term Behavior Editing. Model editing is an emerging area of research that enables precise and efficient modifications to LLMs while preserving their overall capabilities. To systematically study and evaluate this approach, we introduce BehaviorBench, a multi-tier benchmark grounded in psychological moral theories. This benchmark supports both the evaluation and editing of agent behaviors across a variety of scenarios, with each tier introducing more complex and ambiguous scenarios. We first demonstrate that Behavior Editing can dynamically steer agents toward the target behavior within specific scenarios. Moreover, Behavior Editing enables not only scenario-specific local adjustments but also more extensive shifts in an agent’s global moral alignment. We demonstrate that Behavior Editing can be used to promote ethical and benevolent behavior or, conversely, to induce harmful or malicious behavior. Through extensive evaluations of agents built on frontier LLMs, BehaviorBench validates the effectiveness of behavior editing across a wide range of models and scenarios. Our findings offer key insights into a new paradigm for steering agent behavior, highlighting both the promise and perils of Behavior Editing.

The integration of Large Language Models (LLMs) into critical societal and scientific functions has intensified the urgent demand for transparency, reliability, and trust. While post-hoc attribution methods and Chain-of-Thought reasoning currently serve as the dominant approaches to explainability, growing evidence shows that they are often unreliable, producing brittle, misleading, or illusory explanations that fail to reflect true model behavior. This tutorial aims to unpack why these limitations arise. We first establish the theoretical intractability of complete mechanistic explanations for modern LLMs and clarify the intrinsic barriers to achieving full transparency in overparameterized models. We then pivot to a principled alternative: user-centric explainability, with a focus on concept-based interpretability and controlled data attribution. We review the theoretical foundations of these methods and survey their modern extensions that enable comprehensive explanation, inference-time intervention, and model editability. Finally, we demonstrate how such approaches support effective human--AI collaboration in high-stakes scientific and decision-critical applications. By synthesizing foundational theory, critical analysis of existing methods, and emerging techniques, this tutorial offers a coherent framework for developing the next generation of explainable and trustworthy AI systems.

Large Language Models (LLMs) have brought significant advances across various NLP tasks through few-shot or zero-shot prompting, bypassing the need for parameter tuning. However, the "black-box" nature behind their massive parameter sizes increases the "hallucination" concerns, especially in high-stakes applications (e.g., healthcare), where decision mistakes can lead to severe consequences. In contrast, human decision-making relies on complex cognitive processes, such as the ability to sense and adaptively correct mistakes through conceptual understanding. Drawing inspiration from human cognition, we propose an innovative metacognitive approach CLEAR, to equip LLMs with capabilities for self-aware error identification and correction. Our framework constructs concept-specific sparse subnetworks that indicate decision processes. This provides a novel interface for model {intervention} after deployment. The benefits include: (i) at inference time, our metacognitive LLMs can self-consciously identify potential mispredictions with minimum human involvement, (ii) the model can self-correct its errors efficiently without additional tuning, and (iii) the correction procedure is not only self-explanatory but also user-friendly, enhancing model interpretability and accessibility. With these metacognitive features, our approach pioneers a new path toward the trustworthiness of LLMs.

Functional Magnetic Resonance Image (fMRI) is commonly employed to study human brain activity, since it offers insight into the relationship between functional fluctuations and human behavior. To enhance analysis and comprehension of brain activity, Graph Neural Networks (GNNs) have been widely applied to the analysis of functional connectivities (FC) derived from fMRI data, due to their ability to capture the synergistic interactions among brain regions. However, in the human brain, performing complex tasks typically involves the activation of certain pathways, which could be represented as paths across graphs. As such, conventional GNNs struggle to learn from these pathways due to the long-range dependencies of multiple pathways. To address these challenges, we introduce a novel framework BrainMAP to learn multiple pathways in brain networks. BrainMAP leverages sequential models to identify long-range correlations among sequentialized brain regions and incorporates an aggregation module based on Mixture of Experts (MoE) to learn from multiple pathways. Our comprehensive experiments highlight BrainMAP's superior performance. Furthermore, our framework enables explanatory analyses of crucial brain regions involved in tasks.

Large Language Models (LLMs) have made significant progress in open-ended dialogue, yet their inability to retain and retrieve relevant information from long-term interactions limits their effectiveness in applications requiring sustained personalization. External memory mechanisms have been proposed to address this limitation, enabling LLMs to maintain conversational continuity. However, existing approaches struggle with two key challenges. First, rigid memory granularity fails to capture the natural semantic structure of conversations, leading to fragmented and incomplete representations. Second, fixed retrieval mechanisms cannot adapt to diverse dialogue contexts and user interaction patterns. In this work, we propose Reflective Memory Management (RMM), a novel mechanism for long-term dialogue agents, integrating forward- and backward-looking reflections: (1) Prospective Reflection, which dynamically summarizes interactions across granularities—utterances, turns, and sessions—into a personalized memory bank for effective future retrieval, and (2) Retrospective Reflection, which iteratively refines the retrieval in an online reinforcement learning (RL) manner based on LLMs’ cited evidence. Experiments show that RMM demonstrates consistent improvement across various metrics and benchmarks. For example, RMM shows more than 10% accuracy improvement over the baseline without memory management on the LongMemEval dataset.

Most discussions about Large Language Model (LLM) safety have focused on single-agent settings but multi-agent LLM systems now create novel adversarial risks because their behavior depends on communication between agents and decentralized reasoning. In this work, we innovatively focus on attacking pragmatic systems that have constrains such as limited token bandwidth, latency between message delivery, and defense mechanisms. We design a \textit{permutation-invariant adversarial attack} that optimizes prompt distribution across latency and bandwidth-constraint network topologies to bypass distributed safety mechanisms within the system. Formulating the attack path as a problem of \textit{maximum-flow minimum-cost}, coupled with the novel \textit{Permutation-Invariant Evasion Loss (PIEL)}, we leverage {graph-based optimization} to maximize attack success rate while minimizing detection risk. Evaluating across models including \texttt{Llama}, \texttt{Mistral}, \texttt{Gemma}, \texttt{DeepSeek} and other variants on various datasets like \texttt{JailBreakBench} and \texttt{AdversarialBench}, our method outperforms conventional attacks by up to 7\times, exposing critical vulnerabilities in multi-agent systems. Moreover, we demonstrate that existing defenses, including variants of \texttt{Llama-Guard} and \texttt{PromptGuard}, fail to prohibit our attack, emphasizing the urgent need for multi-agent specific safety mechanisms.

Content analysis breaks down complex and unstructured texts into theory-informed numerical categories.Particularly, in social science, this process usually relies on multiple rounds of manual annotation, domain expert discussion, and rule-based refinement.In this paper, we introduce SCALE, 1 a novel multi-agent framework that effectively Simulates Content Analysis via Large language model agEnts.SCALE imitates key phases of content analysis, including text coding, 2 collaborative discussion, and dynamic codebook evolution, capturing the reflective depth and adaptive discussions of human researchers.Furthermore, by integrating diverse modes of human intervention, SCALE is augmented with expert input to further enhance its performance.Extensive evaluations on real-world datasets demonstrate that SCALE achieves human-approximated performance across various complex content analysis tasks, offering an innovative potential for future social science research.* Example: "I love this company's new policy!It's so beneficial for everyone."* Example: "Great job on the recent project!Keep up the good work."-Neutral: Neutral sentiment of users toward the issue/company.* Example: "The company announced a new policy today."* Example: "I heard about the recent changes, but I don't have an opinion yet."

Construction safety research is a critical field in civil engineering, aiming to mitigate risks and prevent injuries through the analysis of site conditions and human factors. However, the limited volume and lack of diversity in existing construction safety datasets pose significant challenges to conducting in-depth analyses. To address this research gap, this paper introduces the Construction Safety Dataset (CSDataset), a well-organized comprehensive multi-level dataset that encompasses incidents, inspections, and violations recorded sourced from the Occupational Safety and Health Administration (OSHA). This dataset uniquely integrates structured attributes with unstructured narratives, facilitating a wide range of approaches driven by machine learning and large language models. We also conduct a preliminary approach benchmarking and various cross-level analyses using our dataset, offering insights to inform and enhance future efforts in construction safety. For example, we found that complaint-driven inspections were associated with a 17.3% reduction in the likelihood of subsequent incidents. Our dataset and code are released at https://github.com/zhenhuiou/Construction-Safety-Dataset-CSDataset.

Graph generation aims to create new graphs that closely align with a target graph distribution. Existing works often implicitly capture this distribution by aligning the output of a generator with each training sample. As such, the overview of the entire distribution is not explicitly captured and used for graph generation. In contrast, in this work, we propose a novel self-conditioned graph generation framework designed to explicitly model graph distributions and employ these distributions to guide the generation process. We first perform self-conditioned modeling to capture the graph distributions by transforming each graph sample into a low-dimensional representation and optimizing a representation generator to create new representations reflective of the learned distribution. Subsequently, we leverage these bootstrapped representations as self-conditioned guidance for the generation process, thereby facilitating the generation of graphs that more accurately reflect the learned distributions. We conduct extensive experiments on generic and molecular graph datasets. Our framework, GraphRCG, demonstrates superior performance over existing state-of-the-art graph generation methods in terms of graph quality and fidelity to training data.

Reward Models (RMs), vital for large model alignment, are underexplored for complex embodied tasks like Embodied Question Answering (EQA) where nuanced evaluation of agents’ spatial, temporal, and logical understanding is critical yet not considerred by generic approaches. We introduce EQA-RM, a novel generative multimodal reward model specifically architected for EQA, trained via our innovative Contrastive Group Relative Policy Optimization (C-GRPO) strategy to learn fine-grained behavioral distinctions. The generative nature of EQA-RM provides interpretable, structured reward feedback (beyond simple scalars), uniquely enabling test-time scaling to dynamically adjust evaluation granularity, from concise scores to detailed critiques of reasoning and grounding, at inference without retraining. Concurrently, we introduce EQARewardBench, a new benchmark built on OpenEQA for standardized EQA reward model assessment. Demonstrating high sample efficiency, EQA-RM (fine-tuning Qwen2-VL-2B-Instruct) achieves 61.9% accuracy on EQA-RM-Bench with 700 samples, outperforming strong proprietary baselines, including Gemini-2.5-Flash, GPT-4o, Claude-3.5-Haiku, and open-sourced state-of-the-art models such as RoVRM and VisualPRM.

Yuhang Chen, Zhen Tan, Ajay Kumar Jaiswal, Huaizhi Qu, Xinyu Zhao, Qi Lin, Yu Cheng, Andrew Kwong, Zhichao Cao, Tianlong Chen. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Advances in large language models (LLMs) significantly enhance reasoning capabilities but their deployment is restricted in resourceconstrained scenarios.Knowledge distillation addresses this by transferring knowledge from powerful teacher models to compact and transparent students.However, effectively capturing the teacher's comprehensive reasoning is challenging due to conventional token-level supervision's limited scope.Using multiple reasoning paths per query alleviates this problem, but treating each path identically is suboptimal as paths vary widely in quality and suitability across tasks and models.We propose Qualityfiltered Routing with Cooperative Distillation (QR-Distill), combining path quality filtering, conditional routing, and cooperative peer teaching.First, quality filtering retains only correct reasoning paths scored by an LLM-based evaluation.Second, conditional routing dynamically assigns paths tailored to each student's current learning state.Finally, cooperative peer teaching enables students to mutually distill diverse insights, addressing knowledge gaps and biases toward specific reasoning styles.Experiments demonstrate QR-Distill's superiority over traditional single-and multi-path distillation methods.Ablation studies further highlight the importance of each component-quality filtering, conditional routing, and peer teaching-in effective knowledge transfer.

Dawei Li, Bohan Jiang, Liangjie Huang, Alimohammad Beigi, Chengshuai Zhao, Zhen Tan, Amrita Bhattacharjee, Yuxuan Jiang, Canyu Chen, Tianhao Wu, Kai Shu, Lu Cheng, Huan Liu. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Real-world vision-language applications demand varying levels of perceptual granularity.However, most existing visual large language models (VLLMs), such as LLaVA, preassume a fixed resolution for downstream tasks, which leads to subpar performance.To address this problem, we first conduct a comprehensive and pioneering investigation into the resolution preferences of different visionlanguage tasks, revealing a correlation between resolution preferences with ❶ image complexity, and ❷ uncertainty variance of the VLLM at different image input resolutions.Building on this insight, we propose an empirical formula to determine the optimal resolution for a given vision-language task, combining these two factors.Second, based on rigorous experiments, we propose a novel parameter-efficient fine-tuning technique to extend the visual input resolution of pre-trained VLLMs to the identified optimal resolution.Extensive experiments on various vision-language tasks validate the effectiveness of our method.

Retrieval-augmented generation (RAG) enhances large language models (LLMs) by integrating external knowledge sources to address their limitations in accessing up-to-date or specialized information.A natural strategy to increase the likelihood of retrieving relevant information is to expand the number of retrieved documents.However, involving more documents could introduce significant noise, as many documents may be irrelevant or misleading, thereby reducing the overall accuracy of the generated responses.To overcome the challenge associated with handling a larger number of documents, we propose WinnowRAG, a novel RAG framework designed to systematically filter out noisy documents while preserving valuable content -a process we refer to as winnowing.WinnowRAG operates in two stages: In Stage I, we perform queryaware clustering to group similar documents and form distinct topic clusters.Each cluster is assigned to an LLM agent for generating a unique answer.In Stage II, we perform winnowing, wherein a critic LLM evaluates the outputs of multiple agents and iteratively separates useful documents from noisy ones.To retain useful documents when discarding agents, we propose two strategic merging techniques to ensure that only relevant knowledge is used for generating the final response.Crucially, WinnowRAG is model-agnostic and does not require any model fine-tuning, making it easily adaptable to various tasks.Extensive experiments on various realistic datasets demonstrate the effectiveness of WinnowRAG over state-ofthe-art baselines.

Recent progress in large language model (LLM)-based multi-agent collaboration highlights the power of structured communication in enabling collective intelligence.However, existing methods largely rely on static or graphbased inter-agent topologies, lacking the potential adaptability and flexibility in communication.In this work, we propose a new framework that rethinks multi-agent coordination through a sequential structure rather than a graph structure, offering a significantly larger topology space for multi-agent communication.Our method focuses on two key directions: (1) Next-Agent Prediction, which selects the most suitable agent role at each step, and (2) Next-Context Selection (NCS), which enables each agent to selectively access relevant information from any previous step.Together, these components construct task-adaptive communication pipelines that support both role flexibility and global information flow.Extensive evaluations across multiple benchmarks demonstrate that our approach achieves superior performance while substantially reducing communication overhead.

As Large Language Models (LLMs) are increasingly deployed to handle various natural language processing (NLP) tasks, concerns regarding the potential negative societal impacts of LLM-generated content have also arisen. To evaluate the biases exhibited by LLMs, researchers have recently proposed a variety of datasets. However, existing bias evaluation efforts often focus on only a particular type of bias and employ inconsistent evaluation metrics, leading to difficulties in comparison across different datasets and LLMs. To address these limitations, we collect a variety of datasets designed for the bias evaluation of LLMs, and further propose CEB, a Compositional Evaluation Bechmark that covers different types of bias across different social groups and tasks. The curation of CEB is based on our newly proposed compositional taxonomy, which characterizes each dataset from three dimensions: bias types, social groups, and tasks. By combining the three dimensions, we develop a comprehensive evaluation strategy for the bias in LLMs. Our experiments demonstrate that the levels of bias vary across these dimensions, thereby providing guidance for the development of specific bias mitigation methods.

In-Context Learning (ICL) empowers Large Language Models (LLMs) to tackle diverse tasks by incorporating multiple input-output examples, known as demonstrations, into the input of LLMs. More recently, advancements in the expanded context windows of LLMs have led to many-shot ICL, which uses hundreds of demonstrations and outperforms few-shot ICL, which relies on fewer examples. However, this approach is often hindered by the high cost of obtaining large amounts of labeled data. To address this challenge, we propose Many-Shot Adaptive Pseudo-LabEling, namely MAPLE, a novel influence-based many-shot ICL framework that utilizes pseudo-labeled samples to compensate for the lack of label information. We first identify a subset of impactful unlabeled samples and perform pseudo-labeling on them by querying LLMs. These pseudo-labeled samples are then adaptively selected and tailored to each test query as input to improve the performance of many-shot ICL, without significant labeling costs. Extensive experiments on real-world datasets demonstrate the effectiveness of our framework, showcasing its ability to enhance LLM adaptability and performance with limited labeled data. Our code is provided at https://github.com/Chen-1031/MAPLE_ICL.

Pretrained language models (PLMs) achieve state-of-the-art results but often function as ``black boxes'', hindering interpretability and responsible deployment. While methods like attention analysis exist, they often lack clarity and intuitiveness. We propose interpreting PLMs through high-level, human-understandable concepts using Concept Bottleneck Models (CBMs). This extended abstract introduces C3M (ChatGPT-guided Concept augmentation with Concept-level Mixup), a novel framework for training Concept-Bottleneck-Enabled PLMs (CBE-PLMs). C3M leverages Large Language Models (LLMs) like ChatGPT to augment concept sets and generate noisy concept labels, combined with a concept-level MixUp mechanism to enhance robustness and effectively learn from both human-annotated and machine-generated concepts. Empirical results show our approach provides intuitive explanations, aids model diagnosis via test-time intervention, and improves the interpretability-utility trade-off, even with limited or noisy concept annotations. This is an concise version of [Tan et al., 2024b], recipient of the Best Paper Award at PAKDD 2024. Code and data are released at https://github.com/Zhen-Tan-dmml/CBM_NLP.git.

With advancements in reasoning capabilities, Large Language Models (LLMs) are increasingly employed for automated judgment tasks. While LLMs-as-Judges offer promise in automating evaluations, current approaches often rely on simplistic aggregation methods (e.g., majority voting), which can fail even when individual agents provide correct answers. To address this, we propose a multi-agent debate judge framework where agents collaboratively reason and iteratively refine their responses. We formalize the debate process mathematically, analyzing agent interactions and proving that debate amplifies correctness compared to static ensembles. To enhance efficiency, we introduce a stability detection mechanism that models judge consensus dynamics via a time-varying Beta-Binomial mixture, with adaptive stopping based on distributional similarity (Kolmogorov-Smirnov test). This mechanism models the judges' collective correct rate dynamics using a time-varying mixture of Beta-Binomial distributions and employs an adaptive stopping criterion based on distributional similarity (Kolmogorov-Smirnov statistic). Experiments across multiple benchmarks and models demonstrate that our framework improves judgment accuracy over majority voting while maintaining computational efficiency.

Existing Embodied Question Answering (EQA) benchmarks primarily focus on household environments, often overlooking safety-critical aspects and reasoning processes pertinent to industrial settings. This drawback limits the evaluation of agent readiness for real-world industrial applications. To bridge this, we introduce IndustryEQA, the first benchmark dedicated to evaluating embodied agent capabilities within safety-critical industrial warehouse scenarios. Built upon the NVIDIA Isaac Sim platform, IndustryEQA provides high-fidelity episodic memory videos featuring diverse industrial assets, dynamic human agents, and carefully designed hazardous situations inspired by real-world safety guidelines. The benchmark includes rich annotations covering six categories: equipment safety, human safety, object recognition, attribute recognition, temporal understanding, and spatial understanding. Besides, it also provides extra reasoning evaluation based on these categories. Specifically, it comprises 971 question-answer pairs generated from small warehouse scenarios and 373 pairs from large ones, incorporating scenarios with and without human. We further propose a comprehensive evaluation framework, including various baseline models, to assess their general perception and reasoning abilities in industrial environments. IndustryEQA aims to steer EQA research towards developing more robust, safety-aware, and practically applicable embodied agents for complex industrial environments.

Large Language Models (LLMs) have achieved unprecedented breakthroughs in various natural language processing domains. However, the enigmatic ``black-box'' nature of LLMs remains a significant challenge for interpretability, hampering transparent and accountable applications. While past approaches, such as attention visualization, pivotal subnetwork extraction, and concept-based analyses, offer some insight, they often focus on either local or global explanations within a single dimension, occasionally falling short in providing comprehensive clarity. In response, we propose a novel methodology anchored in sparsity-guided techniques, aiming to provide a holistic interpretation of LLMs. Our framework, termed SparseCBM, innovatively integrates sparsity to elucidate three intertwined layers of interpretation: input, subnetwork, and concept levels. In addition, the newly introduced dimension of interpretable inference-time intervention facilitates dynamic adjustments to the model during deployment. Through rigorous empirical evaluations on real-world datasets, we demonstrate that SparseCBM delivers a profound understanding of LLM behaviors, setting it apart in both interpreting and ameliorating model inaccuracies. Codes are provided in supplements.

Model attribution for LLM-generated disinformation poses a significant challenge in understanding its origins and mitigating its spread. This task is especially challenging because modern large language models (LLMs) produce disinformation with human-like quality. Additionally, the diversity in prompting methods used to generate disinformation complicates accurate source attribution. These methods introduce domain-specific features that can mask the fundamental characteristics of the models. In this paper, we introduce the concept of model attribution as a domain generalization problem, where each prompting method represents a unique domain. We argue that an effective attribution model must be invariant to these domain-specific features. It should also be proficient in identifying the originating models across all scenarios, reflecting real-world detection challenges. To address this, we introduce a novel approach based on Supervised Contrastive Learning. This method is designed to enhance the model's robustness to variations in prompts and focuses on distinguishing between different source LLMs. We evaluate our model through rigorous experiments involving three common prompting methods: “open-ended”, “rewriting”, and “paraphrasing”, and three advanced LLMs: “llama 2”, “chatgpt”, and “vicuna”. Our results demonstrate the effectiveness of our approach in model attribution tasks, achieving state-of-the-art performance across diverse and unseen datasets.

News media has been frequently utilized as a political tool to stray from facts, making biased statements and claims without evidence. During the COVID-19 vaccine campaign, politically biased news (PBN) has significantly undermined public trust in vaccines. Despite medical evidence showing the benefits of these vaccines, the misperceptions of the vaccine's safety, risks, and efficacy have led to a non-negligible fraction of the population resistant to receiving the vaccine. In this paper, we analyze: (i) how inherent vaccine stances subtly influence individuals' selection of news sources and participation in social media discussions; and (ii) the impact of exposure to PBN on users' attitudes toward vaccines. In doing so, we first curate a comprehensive dataset that connects PBN with related social media discourse. Utilizing advanced deep learning and causal inference techniques, we reveal distinct user behaviors between social media groups with various vaccine stances. Moreover, we observe that individuals with moderate stances, particularly the vaccine-hesitant majority, are more vulnerable to the influence of PBN compared to those with extreme views. Our findings provide critical insights to foster this line of research.

Zhen Tan, Dawei Li, Song Wang, Alimohammad Beigi, Bohan Jiang, Amrita Bhattacharjee, Mansooreh Karami, Jundong Li, Lu Cheng, Huan Liu. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

Retrieval-Augmented Generative (RAG) models enhance Large Language Models (LLMs) by integrating external knowledge bases, improving their performance in applications like fact-checking and information searching.In this paper, we demonstrate a security threat where adversaries can exploit the openness of these knowledge bases by injecting deceptive content into the retrieval database, intentionally changing the model's behavior.This threat is critical as it mirrors real-world usage scenarios where RAG systems interact with publicly accessible knowledge bases, such as web scrapings and user-contributed data pools.To be more realistic, we target a realistic setting where the adversary has no knowledge of users' queries, knowledge base data, and the LLM parameters.We demonstrate that it is possible to exploit the model successfully through crafted content uploads with access to the retriever.Our findings emphasize an urgent need for security measures in the design and deployment of RAG systems to prevent potential manipulation and ensure the integrity of machinegenerated content.

The advent of generative Large Language Models (LLMs) such as ChatGPT has catalyzed transformative advancements across multiple domains. However, alongside these advancements, they have also introduced potential threats. One critical concern is the misuse of LLMs by disinformation spreaders, leveraging these models to generate highly persuasive yet misleading content that challenges the disinformation detection system. This work aims to address this issue by answering three research questions: (1) To what extent can the current disinformation detection technique reliably detect LLM-generated disinformation? (2) If traditional techniques prove less effective, can LLMs themself be exploited to serve as a robust defense against advanced disinformation? and, (3) Should both these strategies falter, what novel approaches can be proposed to counter this burgeoning threat effectively? A holistic exploration for the formation and detection of disinformation is conducted to foster this line of research.

Many text classification methods usually introduce external information (e.g., label descriptions and knowledge bases) to improve the classification performance. Compared to external information, some internal information generated by the model itself during training, like text embeddings and predicted label probability distributions, are exploited poorly when predicting the outcomes of some texts. In this paper, we focus on leveraging this internal information, proposing a dual k nearest neighbor (DkNN) framework with two kNN modules, to retrieve several neighbors from the training set and augment the distribution of labels. For the kNN module, it is easily confused and may cause incorrect predictions when retrieving some nearest neighbors from noisy datasets (datasets with labeling errors) or similar datasets (datasets with similar labels). To address this issue, we also introduce a label distribution learning module that can learn label similarity, and generate a better label distribution to help models distinguish texts more effectively. This module eases model overfitting and improves final classification performance, hence enhancing the quality of the retrieved neighbors by kNN modules during inference. Extensive experiments on the benchmark datasets verify the effectiveness of our method.

Few-shot node classification, which aims to predict labels for nodes on graphs with only limited labeled nodes as references, is of great significance in real-world graph mining tasks. To tackle such a label shortage issue, existing works generally leverage the meta-learning framework, which utilizes a number of episodes to extract transferable knowledge from classes with abundant labeled nodes and generalizes the knowledge to other classes with limited labeled nodes. In essence, the primary aim of few-shot node classification is to learn node embeddings that are generalizable across different classes. To accomplish this, the GNN encoder must be able to distinguish node embeddings between different classes, while also aligning embeddings for nodes in the same class. Thus, in this work, we propose to consider both the intra-class and inter-class generalizability of the model. We create a novel contrastive meta-learning framework on graphs, named COSMIC, with two key designs. First, we propose to enhance the intra-class generalizability by involving a contrastive two-step optimization in each episode to explicitly align node embeddings in the same classes. Second, we strengthen the inter-class generalizability by generating hard node classes for classification via a novel similarity-sensitive mix-up strategy. Extensive experiments on prevalent few-shot node classification datasets verify the effectiveness of our framework and demonstrate its superiority over other state-of-the-art baselines.

The ability to incrementally learn new classes is vital to all real-world artificial intelligence systems. A large portion of high-impact applications like social media, recommendation systems, E-commerce platforms, etc. can be represented by graph models. In this paper, we investigate the challenging yet practical problem,Graph Few-shot Class-incremental (Graph FCL) problem, where the graph model is tasked to classify both newly encountered classes and previously learned classes. Towards that purpose, we put forward a Graph Pseudo Incremental Learning paradigm by sampling tasks recurrently from the base classes, so as to produce an arbitrary number of training episodes for our model to practice the incremental learning skill. Furthermore, we design a Hierarchical-Attention-based Graph Meta-learning framework, HAG-Meta from an optimization perspective. We present a task-sensitive regularizer calculated from task-level attention and node class prototypes to mitigate overfitting onto either novel or base classes. To employ the topological knowledge, we add a node-level attention module to adjust the prototype representation. Our model not only achieves greater stability of old knowledge consolidation, but also acquires advantageous adaptability to new knowledge with very limited data samples. Extensive experiments on three real-world datasets, including Amazon-clothing, Reddit, and DBLP, show that our framework demonstrates remarkable advantages in comparison with the baseline and other related state-of-the-art methods.