Defending against backdoor attacks on graphs has become increasingly critical. Existing methods predominantly focus on detecting and removing triggers by identifying inconsistencies between trigger and clean nodes. However, adversaries can design triggers that closely resemble clean nodes, making them challenging to detect. Therefore, understanding the mechanisms underlying backdoor attacks is crucial. In this work, we observe an interesting phenomenon: in backdoored models, specific ''backdoor neurons'' (embedding dimensions) are more likely to be activated, causing nodes to be misclassified to the target label. This is largely due to the graph structure, where malicious information propagates through node neighborhoods, activating specific neurons and target label. Based on this observation, we theoretically and empirically demonstrate how graph backdoor attacks exploit this propagation mechanism to effectively poison the target node's embedding. Meanwhile, we propose a novel defense called Graph Backdoor Neuron Defense (GBND) to identify, unlearn, and recover backdoor neurons. Specifically, we design a novel reverse engineering technique to identify triggers that activate backdoor neurons, and eliminate their harmful effects by asymmetric unlearning and recovering at the neuron level. Extensive experiments on four datasets validate the effectiveness of GBND in defending against backdoor attacks.
论文检索
输入标题、作者或关键词，从 13,734 篇学术成果中精准定位
Signed Graph Neural Networks (SGNNs) have achieved outstanding performance in Link Sign Prediction (LSP), which involves predicting the existence and polarity of edges, by effectively modeling positive and negative interactions in signed graphs. However, their black-box nature raises transparency concerns, necessitating faithful explanations of model behavior to ensure trustworthiness and accountability. Existing eXplainable GNN (XGNN) methods, primarily designed for unsigned graphs, struggle to provide meaningful and human-understandable explanations for SGNN-based LSP, often generating disconnected subgraph explanations or neglecting the unique sign interactions. To address the gap, we propose SGExplainer, a novel method that leverages balanced paths, a concept rooted in signed graph theory, to provide clear and faithful explanations for LSP. SGExplainer employs a path-enforcing mask learning framework that ensures interpretable balanced path generation while maintaining explanation fidelity. Extensive experiments on real-world signed graphs demonstrate that SGExplainer consistently provides faithful and intuitive explanations for various SGNNs, outperforming state-of-the-art baselines in explanation quality, interpretability, and efficiency.
Foundation models are at the forefront of artificial intelligence. A tokenizer, converting the raw input into discrete representations that the model can understand, plays an important role to the success of foundation models. Unlike the text tokenizer that is well studied in large language models, graph tokenizer is still at its early stage, facing the challenges of tackling the non-Euclidean structures and capturing the structural semantics. How to design a graph tokenizer for structural knowledge transfer? To this end, we propose a Riemannian Graph Tokenizer (RGT ) that bridges the structural knowledge and quantized representations to support cross-domain structural knowledge transfer. The connection is established by Riemannian geometry. Specifically, we first define the geometric vocabulary (trees, cycles and sequences), which captures fundamental structural patterns and reflects the intrinsic geometry of graph. Second, we construct a Riemannian quantizer with Riemannian Straight-Through Estimator to tokenise graph structures across multiple domains into discrete tokens. To ensure consistency and transferability across diverse geometric spaces, RGT further incorporates a geometry-aligned decoder that projects manifold-specific tokens into a unified tangent space. The theoretical analysis and geometric interpretations are provided to support the effectiveness of our proposed method. Extensive experiments across diverse datasets demonstrate that RGT significantly enhances structural knowledge transferability across graph domains.
Extracting cohesive subgraphs from complex networks is a fundamental task in graph analytics and is essential for understanding biological, social, and web graphs. The edge-based γ-quasi-clique model offers a flexible alternative by identifying subgraphs whose edge densities exceed a specified threshold γ. However, finding the exact maximum edge-based quasi-clique is computationally challenging, as the problem is NP-hard and lacks the hereditary property. These characteristics limit the effectiveness of conventional pruning methods and the development of efficient reduction rules. As a result, existing algorithms, such as QClique and FPCE, struggle to scale to large graphs. In this paper, we revisit the problem and propose a novel iterative framework that reformulates the problem as a sequence of hereditary subproblems, enabling more effective pruning and reduction strategies and improving the worst-case time complexity. Furthermore, we redesign the iterative process and introduce a novel heuristic to further improve practical efficiency. Extensive experiments on 253 large-scale real-world graphs demonstrate that our proposed algorithm EQC-Pro outperforms existing methods by up to four orders of magnitude.
Digital labor markets are increasingly used to procure diverse forms of human input, from data annotation to food delivery. A central concern in such markets is the ability of platforms to suppress wages by exploiting the abundance of low-cost labor. To better understand labor outcomes, we introduce a novel posted-price procurement model with coverage objectives. In our model a platform seeks to acquire a set of M tasks while minimizing wait time, as well as total spending. Workers are sampled from a population and complete a task if the posted price is higher than their estimated cost of labor. First, we show that with a simple pricing strategy the platform can cover all categories in time O(M), while paying only a O(log(M)/M) fraction of the total cost of labor by exploiting high worker uncertainty about their costs. Then, we study the impact of collective action to prevent the exploitation of workers. We show how a small, strategically chosen coalition of workers that commits to a price floor forces the platform's total spending from logarithmic to linear in M, substantially reducing the power of the platform to suppress wages. In contrast, a randomly sampled coalition of equal size remains largely ineffective. We complement our theory with synthetic experiments that demonstrate the benefit of targeted recruitment for collective action across different market regimes. More broadly, our results provide a theoretical foundation for understanding how to organize collective bargaining for promoting welfare in digital labor markets.
Online market platforms play an increasingly powerful role in the economy. An empirical phenomenon is that platforms, such as Amazon, Apple, and DoorDash, also enter their own marketplaces, imitating successful products developed by third-party sellers. We formulate a Stackelberg model, where the platform acts as the leader by committing to an entry policy: when will it enter and compete on a product? We study this model through a theoretical and computational framework. We begin with a single seller, and consider different kinds of policies for entry. We characterize the seller's optimal explore-exploit strategy via a Gittins-index policy, and give an algorithm to compute the platform's optimal entry policy. We then consider multiple sellers, to account for competition and information spillover. Here, the Gittins-index characterization fails, and we employ deep reinforcement learning to examine seller equilibrium behavior. Our findings highlight the incentives that drive platform entry and seller innovation, consistent with empirical evidence from markets such as Amazon and Google Play, with implications for regulatory efforts to preserve innovation and market diversity.
The growing scale of ad auctions on online advertising platforms has intensified competition, making manual bidding impractical and necessitating auto-bidding to help advertisers achieve their economic goals. Current auto-bidding methods have evolved to use offline reinforcement learning or generative methods to optimize bidding strategies, but they can sometimes behave counterintuitively due to the black-box training manner and limited mode coverage of datasets, leading to challenges in understanding task status and generalization in dynamic ad environments. Large language models (LLMs) offer a promising solution by leveraging prior human knowledge and reasoning abilities to improve auto-bidding performance. However, directly applying LLMs to auto-bidding faces difficulties due to the need for precise actions in competitive auctions and the lack of specialized auto-bidding knowledge, which can lead to hallucinations and suboptimal decisions. To address these challenges, we propose a hierarchical Large auto-Bidding Model (LBM) to leverage the reasoning capabilities of LLMs for developing a superior auto-bidding strategy. This includes a high-level LBM-Think model for reasoning and a low-level LBM-Act model for action generation. Specifically, we propose a dual embedding mechanism to efficiently fuse two modalities, including language and numerical inputs, for language-guided training of the LBM-Act; then, we propose an offline reinforcement fine-tuning technique termed GQPO for mitigating the LLM-Think's hallucinations and enhancing decision-making performance without simulation or real-world rollout like previous multi-turn LLM-based methods. Experiments demonstrate the superiority of a generative backbone based on our LBM, especially in an efficient training manner and generalization ability.
AI has reached a turning point. Systems can now perceive, generate, and act in language and image across digital platforms at unprecedented scale. Yet as AI moves from tools to collaborators—embedded in decision-making, institutions, and everyday life—a new requirement becomes unavoidable: AI must understand the world the way humans inhabit it. This talk introduces Cognitive World Modeling as the next phase of AI development. It unifies physical world modeling—time, space, causality, action—with mental world modeling—goals, beliefs, intentions, emotions, and social norms—into a single, persistent representation of reality as experienced by humans. Together, these models allow AI systems not only to predict outcomes, but to reason about meaning, context, and consequence. Cognitive World Modeling moves AI beyond reactive toward systems that can plan, explain, adapt, and collaborate over time. Alignment and trust emerge not as post hoc constraints, but as properties of systems that maintain accurate, evolving models of both the external world and the humans within it.
This demo paper describes the development of the AI Teaching & Learning Assistant, a modular Moodle plugin that leverages Retrieval-Augmented Generation (RAG) to deliver high-quality, highly selective answers strictly grounded in teachers’ materials. The system employs a dual-centric design, providing students with interactive, Socratic-based tutoring options and educators with a “human-in-the-loop” workspace for supervised content generation, e.g., quizzes, tasks, or short summaries. By grounding Large Language Model (LLM) responses in teacher-provided materials, the assistant addresses the risks of misinformation while encouraging deep conceptual mastery. Evaluation via the LLM-as-a-Judge framework and a preliminary user study confirms its effectiveness, achieving faithfulness scores up to 0.97 and a 4.00/5.00 recommendation rate. Demo Video: https://tinyurl.com/4zz4bcjn
As AI‑based decision‑support systems become increasingly widespread, methods aimed at improving the performance of human-AI teams are gaining attention. In recent years, explainable artificial intelligence (XAI) has received growing interest, as it provides methods to make the behavior of machine learning models more transparent and can help to identify errors and flaws, which is particularly important in safety critical domains such as medicine and law. However, explanations themselves can be misleading, inconsistent, or incorrect, which makes it essential to raise awareness of the possibilities and limitations of these methods. We introduce Lucyde, a web‑based demonstrator designed to help users explore, compare, and better understand XAI methods across different datasets, models, and configurations. Lucyde provides a curated collection of explanation techniques, enables side‑by‑side comparison of methods, and offers easy‑to‑understand supplementary information for different user groups. It also illustrates interactive machine learning workflows by allowing users to correct model outputs or explanations. Lucyde thereby fosters informed and reflective engagement with AI systems and their explanations.
While Open-Set Semantic Mapping and 3D Semantic Scene Graphs (3DSSGs) have become established paradigms in robotic perception in recent years, most existing works are limited to small environments or sacrifice geometric detail and instance granularity for scalability. Deploying these systems at scale for large multi-room environments remains a major challenge due to the computational overhead of high-dimensional feature integration and the maintenance of the 3DSSG structure. In this paper, we demonstrate a modular mapping architecture that establishes 3D Semantic Scene Graphs (3DSSGs) as its foundational backend. Unlike approaches that generate scene graphs as a post-processing step, our system maintains the graph as the primary, incrementally updated knowledge representation. Our architecture is optimized for GPU-accelerated operations, enabling the dense representation of extensive environments containing thousands of unique object instances, supporting open-vocabulary queries via CLIP features without requiring any additional post-processing steps. In this live demonstration, we showcase our pipeline processing large-scale data from the Habitat Matterport 3D (HM3D) dataset as well as live data collected from a handheld device. Attendees will interact with the generated maps by performing real-time, open-set queries (e.g., “find the vintage wooden chair”) across complex, multi-story environments, highlights the system's capability to represent dynamic, human-aligned environmental understanding suitable for downstream robotic tasks.
Active learning reduces annotation effort by selecting informative samples, yet most approaches remain model-driven, offering users little control over training or support for understanding model behaviour. Human-centric active learning brings users further into the loop by introducing additional points of interaction, particularly in the sample selection process. However, such systems are typically demonstrated using fixed feature projections or visualizations of shallow classifier outputs. We present a representation-centric active learning tool in which interaction takes place directly within the model’s representation space. By operating in the same space the model uses for decision making, the interface supports the co-evolution of representations and user understanding. We additionally report initial qualitative (think-aloud) and quantitative findings from a pilot study, illustrating that such representation-centric frameworks can achieve comparable performance to standard baselines while fostering improved human–model collaboration. (Video and the code available at https://cst.dfki.de/demo-interacting-model-space)
Large-scale video repositories are increasingly available for modern video understanding and generation tasks. However, transforming raw videos into high-quality, task-specific datasets remains costly and inefficient. We present DataCube, an intelligent platform for automatic video processing, multi-dimensional profiling, and query-driven retrieval. DataCube constructs structured semantic representations of video clips and supports hybrid retrieval with neural re-ranking and deep semantic matching. Through an interactive web interface, users can efficiently construct customized video subsets from massive repositories for training, analysis, and evaluation, and build searchable systems over their own private video collections. The system is publicly accessible at https://datacube.baai.ac.cn/. Demo Video: https://youtu.be/L7bKfPBm2tU
We demonstrate ADP-MA (Autonomous Data Processing using Meta-Agents), a system that autonomously solves a complex and diverse set of data processing tasks. Three domain-agnostic meta-agents coordinate task-specific ground agents through a multi-stage pipeline: data understanding, planning, critique, expansion, execution, and finalization. Errors are caught early via progressive sampling on small data subsets before running on full data. The system supports three execution strategies, twelve domain knowledge packs, and confidence-based early stopping. An interactive web interface lets users watch pipelines being built in real time, replay completed runs at any stage, and compare results across cases. On four benchmarks, ADP-MA reaches 90.6% on DSEval, 44.8% on KramaBench, 50.0% on DA-Code, and 70.0% on AgentBench, outperforming published single-agent baselines.
AI's increasing role as a personal agent assisting knowledge workers in everyday tasks underscores the need to investigate how to help human–agent teams build a shared understanding. We extend the collaborative "mind-reading" game Wavelength to include an AI teammate, presenting the first demonstration of an LLM capable of playing this game. Based on our agent–agent play experiments, we developed Wavelength.AI, which implements two strategies to support shared understanding: an initial team grounding conversation and post-game reflective explanations. We interpret higher team scores as evidence for better shared understanding in a preliminary user study with 24 human–AI teams. Our findings reveal that Wavelength.AI can help researchers evaluate and design different strategies to shape human-agent teams' shared understanding. Human players can see if they are on the same wavelength with AI and view our demo video today at https://play-wavelength-ai.com.