Searcharxiv⌕ Search

arXiv subjects

Yilong Wang

Publications and source records attributed to Yilong Wang.

At least 19 recordsLinked to original sources

Multi-Granularity Context-Enhanced RAG over Multimodal Knowledge Graphs

Knowledge graph (KG)-based retrieval-augmented generation (RAG) leverages structured knowledge to provide large language models (LLMs) with high-quality external information. Building on these works, recent studies have explored multimodal knowledge graphs (MMKGs) as knowledge bases for GraphRAG. This enables GraphRAG to integrate knowledge across multiple modalities, thereby further enhancing its performance. However, existing MMKG-based RAG methods generally follow a common pipeline in which different modalities are largely processed independently before being fused. As a result, textual context is only used to a limited extent during visual information extraction and subsequent multimodal knowledge fusion. This brings a semantic gap between images and text which limits the multimodal GraphRAG performance. To address this issue, we propose a novel framework for constructing a Context-Enhanced MMKG (CEMMKG) to better support multimodal GraphRAG. The proposed CEMMKG enriches each image with complementary textual context at both local and global scopes. Local context goes beyond the surrounding text by incorporating sentences that are semantically related to the image, while global context provides a summary of the entire document. We further introduce a multi-granularity design for the local context, allowing it to capture semantically relevant information at different levels of detail. Extensive experiments on multiple datasets validate that CEMMKG is effective in leveraging contextual information to improve MMKG-based RAG performance. Moreover, its effectiveness across different MMKG-based RAG methods demonstrates its broad applicability.

cs.AI↗

Stretched near-groups

We study stretched near-group fusion categories over $\mathbb{C}$ of type $G+k|G|$ for abelian groups $G$ and positive integers $k$, describing the structure of the double including fusion rules, modular data, and Frobenius--Schur indicators under a variety of assumptions. For cyclic $G$, when enough invertible objects exist in the double, we describe all associated de-equivariantizations which connect the existence of stretched near-group fusion categories to the existence of other stretched quadratic categories. All of our results apply to the case $k=1$. In particular, the known existence of $C_{m^2}+m^2$ near-group categories for all positive integers $m$ implies the existence of Haagerup--Izumi fusion categories for all cyclic groups. As a further application, we provide a general conductor bound severely restricting the existence of stretched near-group categories, and for infinitely many abelian groups $G$ of square order, among them $C_4\times C_2^{2j}$ with $j\geq2$, we show that a near-group fusion category $G+k|G|$ exists if and only if $k=0$.

math.QA↗

Reverse Module Categories of Vertex Operator Algebras

We study the realization of the reverse braided category of the module category of a rational vertex operator algebra V. Using holomorphic extensions and commutant constructions, we prove that this realization property is preserved under suitable cosets, tensor products, extensions, and finite-group orbifolds. As applications, we establish the property for unitary Virasoro minimal models, affine vertex operator algebras of types ABCDE at positive integral levels, the corresponding principal W-algebras of type ADE, and related parafermion vertex operator algebras.

math.QA↗

IXPE observation of the new magnetar source EP J223759.5+531421

We report on the detection of polarized X-ray emission from the new Galactic magnetar EP J223759.5+531421, discovered by the Einstein Probe Wide-field X-ray Telescope on 2026 June 28. The Imaging X-ray Polarimetry Explorer (IXPE) follow-up observation started on 2026 July 6 and detected the source at a (absorbed) flux of $\approx 4.5\times 10^{-11} \, \mathrm{ erg\,cm^2\,s}^{-1}$ ($2$--$7.5\, \mathrm{keV}$ range), confirming the presence of two thermally emitting regions on the star surface, with temperatures $\approx 0.5$ and $\approx 1\, \mathrm{keV}$. A significant (at $> 4σ$ confidence level) phase- and energy-integrated polarization degree of $\approx 7\%$ was detected with the polarization angle $\approx -2^\circ$ East of the celestial North. The two thermal components exhibit quite different polarization properties, with the hotter one being more polarized. The difference is larger in the phase-folded data, with the polarization degree reaching $\sim 63\%$ in the range $4$--$7.5\,\mathrm{keV}$, in correspondence with the secondary peak of the pulse profile, and never exceeding $\sim 30\%$ at lower energies. The polarization angle continuously oscillates from $-90^\circ$ to $+90^\circ$ over one rotational cycle and is well fit by the rotating vector model, with an inclination of the line of sight and of the magnetic dipole axis relative to the star spin axis of $\approx 28^\circ$ and $\approx 109^\circ$, respectively. The data point to an emission geometry in which the cold thermal component likely originates from two antipodal caps, possibly in a magnetically condensed state, while the hotter one comes from a smaller region covered by an atmosphere.

astro-ph.HE↗

MMKG-R1: Constructing Multimodal Knowledge Graphs via Reinforcement Learning

Multimodal knowledge graphs (MMKGs) organize information across different modalities to support multimodal Graph Retrieval-Augmented Generation (RAG), which aims to improve the performance of multimodal large language models (MLLMs). The effectiveness of these multimodal Graph RAG methods largely depends on the quality of the constructed MMKGs. However, most existing methods construct MMKGs without explicit guidance on graph quality, while reinforcement learning (RL)-based methods that rely on final-answer correctness may provide only indirect feedback on MMKG quality. To address these limitations, we propose MMKG-R1, a novel MMKG-quality-driven RL framework for training an MLLM as a multimodal knowledge constructor. To be specific, MMKG-R1 introduces multiple quality-aware rewards to directly evaluate both the quality of information extracted from individual knowledge sources and the quality of the assembled MMKG. These rewards are then used to guide the RL training. Experiments across multiple datasets show that MMKG-R1 constructs MMKGs that improve downstream question-answering accuracy.

cs.CV↗

Instant-Fold: In-Context Imitation Learning for Deformable Object Manipulation

Deformable object manipulation (DOM) is challenging due to high-dimensional, partially observable states that evolve through long-horizon, topology-changing interactions with multiple valid manipulation modes. We introduce Instant-Fold, an in-context imitation learning framework for DOM. Given a single human demonstration, our policy infers and executes diverse manipulation modes directly from the demonstration, including variations in spatial execution and ordering, without requiring gradient updates. Our approach first learns deformation-aware visual representations via temporal contrastive pretraining, after which a flow-matching transformer policy conditioned on the demonstration predicts actions to execute the intended manipulation mode. Trained entirely in simulation, Instant-Fold generalizes across diverse folding modes and transfers zero-shot to real-world settings without additional data collection or finetuning. Videos are available at https://instant-fold.github.io.

cs.RO↗

Modular fusion categories with trivial Torelli group actions

In this paper, we study modular fusion categories whose mapping class group representations are trivial on the Torelli groups, with particular emphasis on the congruence properties of the resulting representations of $\mathrm{Sp}(2g, \mathbb{Z})$. We prove that, for any $g \ge 3$, the Torelli group is contained in the kernel of the genus-$g$ mapping class group representation $ρ_g$ associated with a modular fusion category $\mathcal{C}$ if and only if $\mathcal{C}$ is pointed. This refines the corresponding result in \cite{MW25}. We also characterize the modular fusion categories for which the genus-$2$ Torelli group is contained in $\ker ρ_2$; in particular, categories with isotropic adjoint subcategories belong to this class. Finally, we give a complete classification of modular fusion categories with isotropic adjoint subcategories, and determine the groups $G$ satisfying $\mathcal{Z}(\mathrm{Vect}_G)$ is 2-Torelli trivial but $\mathcal{Z}(\mathrm{Vect}_G)_{\mathrm{ad}}$ is not isotropic are precisely 2-Engel groups of nilpotent class 3.

math.QA↗

From Concept Alignment to Causal Grounding: An Intervention Test of Chain-of-Thought Faithfulness

Chain-of-thought (CoT) can sound plausible yet be unfaithful to the model's underlying reasoning. Most prior work probes CoT faithfulness through input--output behavior or input attributions, leaving internal computation largely underexplored. We instead cast faithfulness as internal concept grounding: Does a large language model's (LLM) CoT reasoning engage the same internal concepts that support the LLM's direct prediction, and do the shared concepts causally drive its answer? Encoding a prediction pass and a CoT pass with a single shared sparse autoencoder (SAE), a reliable approximator of the latent concepts LLMs use, makes their internal concepts directly comparable. We introduce three correlational metrics of concept-level alignment and a causal metric, $Δp$, which ablates the shared concepts and measures the drop in answer probability. Across five LLMs and four datasets, concept alignment is generally high, as indicated by the correlational metrics; yet these only identify which concepts are shared, not how much they causally contribute. $Δp$ fills this gap: causal faithfulness varies substantially with model depth, peaking at mid-to-late layers rather than the final ones, and model scale reshapes the layer-wise profile. Moreover, causally important shared concepts are not always verbalized in the CoT. These dissociations suggest that faithfulness cannot be reliably assessed from surface-level or representational correspondence alone; assessing it requires causal tests of whether the internal concepts underlying a CoT actually drive the model's prediction.

cs.CL↗

Low latency global carbon budget reveals strong land sink recovery in 2025

The atmospheric CO2 growth rate fell sharply in 2025, from a record 3.76 $\pm$ 0.09 ppm yr-1 in 2024 to 2.06 $\pm$ 0.09 ppm yr-1 (NOAA marine boundary layer observations), below the 2015-2022 mean of 2.47 ppm yr-1, even as fossil CO2 emissions rose by 0.7% to 10.38 GtC yr-1. Here we present a low-latency global and regional carbon budget for 2025, combining three dynamic global vegetation models (DGVMs) and ocean model emulators with four atmospheric inversions constrained by OCO-2 satellite retrievals. The global net land sink reached 2.36 $\pm$ 0.16 GtC yr-1 in 2025 (DGVMs: 2.04 $\pm$ 0.24; inversions: 2.68 $\pm$ 0.20 GtC yr-1), strengthening by 2.81 $\pm$ 0.31 GtC yr-1 from 2024 and exceeding the 2015-2022 mean by 0.71 $\pm$ 0.13 GtC yr-1. Ocean uptake (3.11 $\pm$ 0.36 GtC yr-1) remained similar to 2024, making the land sink rebound the dominant driver of the slowdown in CO2 growth. Tropical lands shifted from net sources in 2024 to net sinks in 2025, with enhanced uptake across much of Africa and northern Eurasia, and land flux anomalies covaried with GRACE terrestrial water storage. Where the sink had weakened substantially in 2023-2024, about 80% of the area showed some recovery, with overall recovery of 87.3% (DGVMs) to 99.5% (inversions). Recovery exceeded 100% in the tropics but remained incomplete in the northern extratropics, indicating a strong but spatially uneven rebound of the land carbon sink.

physics.ao-ph↗

iFlip: Iterative Feedback-driven Counterfactual Example Refinement

Counterfactual examples are minimal edits to an input that alter a model's prediction. They are widely employed in explainable AI to probe model behavior and in natural language processing (NLP) to augment training data. However, generating valid counterfactuals with large language models (LLMs) remains challenging, as existing single-pass methods often fail to induce reliable label changes, neglecting LLMs' self-correction capabilities. To explore this untapped potential, we propose iFlip, an iterative refinement approach that leverages three types of feedback, including model confidence, feature attribution, and natural language. Our results show that iFlip achieves an average 57.8% higher validity than the five state-of-the-art baselines, as measured by the label flipping rate. The user study further corroborates that iFlip outperforms baselines in completeness, overall satisfaction, and feasibility. In addition, ablation studies demonstrate that three components are paramount for iFlip to generate valid counterfactuals: leveraging an appropriate number of iterations, pointing to highly attributed words, and early stopping. Finally, counterfactuals generated by iFlip enable effective counterfactual data augmentation, substantially improving model performance and robustness.

cs.CL↗

Macro: Enhancing Multilingual Counterfactual Explanations through Alignment-as-Preference Optimization

Self-generated counterfactual explanations (SCEs) are minimally modified inputs (minimality) generated by large language models (LLMs) that flip their own predictions (validity), offering a causally grounded approach to unraveling black-box LLM behavior. Yet extending them beyond English remains challenging: existing methods struggle to produce valid SCEs in non-dominant languages, and a persistent trade-off between validity and minimality undermines explanation quality. We introduce Macro, a preference alignment framework that applies Direct Preference Optimization (DPO) to multilingual SCE generation, using a composite scoring function to construct preference pairs that effectively translate the trade-off into measurable preference signals. Experiments across four LLMs and seven typologically diverse languages show that Macro improves validity by 12.55\% on average over the chain-of-thought baseline without degrading minimality, while avoiding the severe minimality violations of the translation-based baseline. Compared to supervised fine-tuning, Macro achieves superior performance on both metrics, confirming that explicit preference optimization is essential for balancing this trade-off. Further analyses reveal that Macro increases cross-lingual perturbation alignment and mitigates common generation errors. Our results highlight preference optimization as a promising direction for enhancing multilingual model explanations.

cs.CL↗

Trojaning the Alignment: Stealthy Backdoor Attacks against Graph Foundation Models

Graph Foundation Models (GFMs) on text-attributed graphs (TAGs) align graph representations with language semantics to support transferable graph learning. Despite these advantages, the backdoor vulnerability of GFMs on TAGs remains insufficiently understood, especially under graph-language alignment, where graph and text representations are trained to constrain each other in a shared semantic space. Existing backdoor attacks mainly target either the graph side or the text side, treating the two modalities independently. This makes direct adaptation ineffective: graph-only triggers can be constrained by clean text semantics, while text-only triggers alter the language view but do not directly shift the graph representation being aligned and scored. TAGs also impose a stealth challenge because triggers are exposed as both node text and local graph structure, making incoherent trigger attributes or anomalous subgraphs easy to inspect or filter. In this paper, we propose STAG, a stealthy trojan attack framework designed for the graph-language alignment interface of GFMs on TAGs. STAG coordinates a graph-trigger generator with a text-side soft prompt so that trigger-attached graph representations and triggered text representations move toward the same target-class text region. To address TAG-specific stealthiness, STAG realizes trigger nodes as readable text through candidate retrieval and regularizes the trigger-attached subgraph so that its local structure remains close to the original subgraph. Extensive experiments on multiple TAG datasets and representative GFMs demonstrate the effectiveness and stealthiness of STAG. Our code is available at https://github.com/ventr1c/STAG.

cs.LG↗

Non-semisimple open-closed 3d TFT

Given a spherical finite tensor category C, not necessarily semisimple, and a two-sided modified trace on its projective ideal, we define an open-closed three-dimensional topological field theory with values in vector spaces. The bordism category has as morphisms three-dimensional bordisms with corners, whose boundary is partitioned into the gluing boundary, parametrised by source and target surface, and the unparametrised free boundary. The free boundary is equipped with an embedded C-coloured graph satisfying an admissibility condition. Our construction starts from a new three-manifold invariant based on the multi-handlebody invariant of [arXiv:1809.07991] and the chromatic maps of [arXiv:2302.04509]. The open-closed topological field theory is then obtained via the universal construction.

math.QA↗

Non-invasive Blood Glucose Estimation from Wearable Physiological Signals

Non-invasive blood glucose estimation from wearable physiological signals remains difficult because longitudinal photoplethysmography (PPG) data are subject to distribution drift, whereas reference capillary blood glucose labels are sparse and costly to acquire. We propose a \rev{deep-learning-based} dynamic incremental learning (DIL) framework that combines a mutual entropy-optimized replay-based dynamic clustering module (MERDC) with an uncertainty-quantified proxy gradient bridging agent (PGBA) for label-efficient adaptation to unlabeled PPG streams. To support this setting, we further establish a longitudinal benchmark dataset comprising PPG, reference capillary blood glucose, and cuff blood pressure measurements from 183 participants collected over 285 days, and we make this resource available to the research community. Under 5-fold subject-independent validation, the proposed method achieves a mean absolute error (MAE) of $0.64 \pm 0.01$ millimoles per liter (mmol/L) and a root mean square error (RMSE) of $1.29 \pm 0.10$ mmol/L, with $97.69 \pm 1.63\%$ of estimates falling within Clarke zones A+B. Aggregation-level analyses further support the robustness of the observed error distribution beyond window-level evaluation. \rev{These results provide a proof-of-concept for adaptive non-invasive glucose estimation in wearable physiological sensing and establish a longitudinal benchmark for subsequent research.

eess.SP↗

SynthICL: Scalable In-context Imitation Learning with Synthetic Data

In-context imitation learning (ICIL) enables robots to learn new tasks from a small number of demonstrations by conditioning a pre-trained policy on task-specific examples, without retraining at test time. Despite this promise, training generalizable and scalable in-context imitation policies remains an open challenge. We present SynthICL, a scalable framework that trains ICIL policies entirely from RGB-only synthetic data. Specifically, we build a data generation pipeline to produce high-fidelity ICIL data and train a flow-matching transformer policy on the resulting dataset. SynthICL avoids the need for depth sensing, precise camera calibration, and real-world training data in prior approaches, offering a simpler and more scalable alternative. We further incorporate subgoal prediction by training the model to predict the next subgoal images, enabling more precise and visually grounded control. Evaluated on 16 unseen real-world manipulation tasks, SynthICL achieves an average success rate of 79% with only one demonstration provided at test time and outperforms prior methods. Project page: https://synth-icl.github.io

cs.RO↗

Query-Efficient Agentic Graph Extraction Attacks on GraphRAG Systems

Graph-based retrieval-augmented generation (GraphRAG) systems construct knowledge graphs over document collections to support multi-hop reasoning. While prior work shows that GraphRAG responses may leak retrieved subgraphs, the feasibility of query-efficient reconstruction of the hidden graph structure remains unexplored under realistic query budgets. We study a budget-constrained black-box setting where an adversary adaptively queries the system to steal its latent entity-relation graph. We propose AGEA (Agentic Graph Extraction Attack), a framework that leverages a novelty-guided exploration-exploitation strategy, external graph memory modules, and a two-stage graph extraction pipeline combining lightweight discovery with LLM-based filtering. We evaluate AGEA on medical, agriculture, and literary datasets across Microsoft-GraphRAG and LightRAG systems. Under identical query budgets, AGEA significantly outperforms prior attack baselines, recovering up to 90% of entities and relationships while maintaining high precision. These results demonstrate that modern GraphRAG systems are highly vulnerable to structured, agentic extraction attacks, even under strict query limits. The code is available at https://github.com/shuashua0608/AGEA.

cs.AI↗

Attack by Unlearning: Unlearning-Induced Adversarial Attacks on Graph Neural Networks

Graph neural networks (GNNs) are widely used for learning from graph-structured data in domains such as social networks, recommender systems, and financial platforms. To comply with privacy regulations like the GDPR, CCPA, and PIPEDA, approximate graph unlearning, which aims to remove the influence of specific data points from trained models without full retraining, has become an increasingly important component of trustworthy graph learning. However, approximate unlearning often incurs subtle performance degradation, which may incur negative and unintended side effects. In this work, we show that such degradations can be amplified into adversarial attacks. We introduce the notion of \textbf{unlearning corruption attacks}, where an adversary injects carefully chosen nodes into the training graph and later requests their deletion. Because deletion requests are legally mandated and cannot be denied, this attack surface is both unavoidable and stealthy: the model performs normally during training, but accuracy collapses only after unlearning is applied. Technically, we formulate this attack as a bi-level optimization problem: to overcome the challenges of black-box unlearning and label scarcity, we approximate the unlearning process via gradient-based updates and employ a surrogate model to generate pseudo-labels for the optimization. Extensive experiments across benchmarks and unlearning algorithms demonstrate that small, carefully designed unlearning requests can induce significant accuracy degradation, raising urgent concerns about the robustness of GNN unlearning under real-world regulatory demands. The source code will be released upon paper acceptance.

cs.LG↗

Observer-Actor: Active Vision Imitation Learning with Sparse-View Gaussian Splatting

We propose Observer Actor (ObAct), a novel framework for active vision imitation learning in which the observer moves to optimal visual observations for the actor. We study ObAct on a dual-arm robotic system equipped with wrist-mounted cameras. At test time, ObAct dynamically assigns observer and actor roles: the observer arm constructs a 3D Gaussian Splatting (3DGS) representation from three images, virtually explores this to find an optimal camera pose, then moves to this pose; the actor arm then executes a policy using the observer's observations. This formulation enhances the clarity and visibility of both the object and the gripper in the policy's observations. As a result, we enable the training of ambidextrous policies on observations that remain closer to the occlusion-free training distribution, leading to more robust policies. We study this formulation with two existing imitation learning methods -- trajectory transfer and behavior cloning -- and experiments show that ObAct significantly outperforms static-camera setups: trajectory transfer improves by 145% without occlusion and 233% with occlusion, while behavior cloning improves by 75% and 143%, respectively. Videos are available at https://obact.github.io.

cs.RO↗