arXiv subjects

Xi Zhang

Publications and source records attributed to Xi Zhang.

At least 73 records · Page 4Linked to original sources

arXiv2026

Uncertainty Quantification on Graph Learning: A Survey

Graphical models have demonstrated their exceptional capabilities across numerous applications. However, their performance, confidence, and trustworthiness are often limited by the inherent randomness in data generation and the lack of knowledge to accurately model real-world complexities. There has been increased interest in developing uncertainty quantification (UQ) techniques tailored to graphical models. In this survey, we systematically examine existing works on UQ for graphical models. This survey distinguishes itself from most existing UQ surveys by specifically concentrating on graphical models, including graph neural networks and graph foundation models. We organize the literature along two complementary dimensions: uncertainty representation and uncertainty handling. By synthesizing both established methodologies and emerging trends, we aim to bridge gaps in understanding key challenges and opportunities in UQ for graphical models, inspiring researchers on graphical models or uncertainty quantification to make further advancements at the cross of the two fields.

cs.LG↗

arXiv2026

Normalized solutions for a class of fractional Choquard equations with the HLS lower critical term and a nonlocal perturbation

In this paper, we study the mass-constrained fractional Choquard equation $ (-Δ)^s u = λu + α(I_μ* |u|^{\frac{2N-μ}{N}})|u|^{\frac{2N-μ}{N}-2}u + (I_μ* |u|^p)|u|^{p-2}u $ in $ \mathbb{R}^N $, under the constraint $ \int_{\mathbb{R}^N} |u|^2 \, dx = c^2 > 0 $, where $ N > 2s $, $ s \in (0,1) $, $ μ\in (0,N) $, $ α> 0 $, and $ 2 + \frac{2s-μ}{N} \le p < \frac{2N-μ}{N-2s} $. We first establish a nonexistence result in the $ L^2 $-critical case $ p = 2 + \frac{2s-μ}{N} $. Then, in the $ L^2 $-supercritical range, we prove the existence of normalized ground states in two complementary regimes determined by the quantity $ \mathcal{M}_1(c) $. Our approach is based on constrained variational methods, a min-max construction, and refined estimates for the associated fiber maps.

math.AP↗

arXiv2026

Deep Light Pollution Removal in Night Cityscape Photographs

Nighttime photography is severely degraded by light pollution induced by pervasive artificial lighting in urban environments. After long-range scattering and spatial diffusion, unwanted artificial light overwhelms natural night luminance, generates skyglow that washes out the view of stars and celestial objects and produces halos and glow artifacts around light sources. Unlike nighttime dehazing, which aims to improve detail legibility through thick air, the objective of light pollution removal is to restore the pristine night appearance by neutralizing the radiative footprint of ground lighting. In this paper we introduce a physically-based degradation model that adds to the previous ones for nighttime dehazing two critical aspects; (i) anisotropic spread of directional light sources, and (ii) skyglow caused by invisible surface lights behind skylines. In addition, we construct a training strategy that leverages large generative model and synthetic-real coupling to compensate for the scarcity of paired real data and enhance generalization. Extensive experiments demonstrate that the proposed formulation and learning framework substantially reduce light pollution artifacts and better recover authentic night imagery than prior nighttime restoration methods.

cs.CV↗

arXiv2026

BADiff: Bandwidth Adaptive Diffusion Model

In this work, we propose a novel framework to enable diffusion models to adapt their generation quality based on real-time network bandwidth constraints. Traditional diffusion models produce high-fidelity images by performing a fixed number of denoising steps, regardless of downstream transmission limitations. However, in practical cloud-to-device scenarios, limited bandwidth often necessitates heavy compression, leading to loss of fine textures and wasted computation. To address this, we introduce a joint end-to-end training strategy where the diffusion model is conditioned on a target quality level derived from the available bandwidth. During training, the model learns to adaptively modulate the denoising process, enabling early-stop sampling that maintains perceptual quality appropriate to the target transmission condition. Our method requires minimal architectural changes and leverages a lightweight quality embedding to guide the denoising trajectory. Experimental results demonstrate that our approach significantly improves the visual fidelity of bandwidth-adapted generations compared to naive early-stopping, offering a promising solution for efficient image delivery in bandwidth-constrained environments. Code is available at: https://github.com/xzhang9308/BADiff.

cs.CV↗

arXiv2026

General Multimodal Protein Design Enables DNA-Encoding of Chemistry

Evolution is an extraordinary engine for enzymatic diversity, yet the chemistry it has explored remains a narrow slice of what DNA can encode. Deep generative models can design new proteins that bind ligands, but none have created enzymes without pre-specifying catalytic residues. We introduce DISCO (DIffusion for Sequence-structure CO-design), a multimodal model that co-designs protein sequence and 3D structure around arbitrary biomolecules, as well as inference-time scaling methods that optimize objectives across both modalities. Conditioned solely on reactive intermediates, DISCO designs diverse heme enzymes with novel active-site geometries. These enzymes catalyze new-to-nature carbene-transfer reactions, including alkene cyclopropanation, spirocyclopropanation, B-H, and C(sp$^3$)-H insertions, with high activities exceeding those of engineered enzymes. Random mutagenesis of a selected design further confirmed that enzyme activity can be improved through directed evolution. By providing a scalable route to evolvable enzymes, DISCO broadens the potential scope of genetically encodable transformations. Code is available at https://github.com/DISCO-design/DISCO.

cs.LG↗

arXiv2026

Beyond Individual Mimicry: Constructing Human-Like Social network with Graph-Augmented LLM Agents

Driven by large language models (LLMs), social bot can autonomously engage in local interactions, whose human-like behaviors enable them to evade social bot detection. However, while these botnets exhibit realistic local social interactions, they fail to preserve human-like social network. This is because LLM-based bots are graph-unaware and cannot coordinate over global interactions, which makes those botnets vulnerable to graph neural network (GNN)-based detection. To address this limitation, we propose GraphMind, which equips LLM-driven social bots to explicitly learn and fit human-like social network structures. Building on this foundation, we further construct GraphMind-Botnet, a LLM-driven botnet designed to evaluate the performance of existing social bot detection algorithms. Experiments on datasets derived from GraphMind-Botnet show that both text-based and graph-based detection models show substantially degraded performance in distinguishing. Our results highlight the critical role of social link construction in LLM-driven social network generation, while exposing fundamental weaknesses in existing bot detection mechanisms.

cs.SI↗

arXiv2026

ClawKeeper: Comprehensive Safety Protection for OpenClaw Agents Through Skills, Plugins, and Watchers

OpenClaw has rapidly established itself as a leading open-source autonomous agent runtime, offering powerful capabilities including tool integration, local file access, and shell command execution. However, these broad operational privileges introduce critical security vulnerabilities, transforming model errors into tangible system-level threats such as sensitive data leakage, privilege escalation, and malicious third-party skill execution. Existing security measures for the OpenClaw ecosystem remain highly fragmented, addressing only isolated stages of the agent lifecycle rather than providing holistic protection. To bridge this gap, we present ClawKeeper, a real-time security framework that integrates multi-dimensional protection mechanisms across three complementary architectural layers. (1) \textbf{Skill-based protection} operates at the instruction level, injecting structured security policies directly into the agent context to enforce environment-specific constraints and cross-platform boundaries. (2) \textbf{Plugin-based protection} serves as an internal runtime enforcer, providing configuration hardening, proactive threat detection, and continuous behavioral monitoring throughout the execution pipeline. (3) \textbf{Watcher-based protection} introduces a novel, decoupled system-level security middleware that continuously verifies agent state evolution. It enables real-time execution intervention without coupling to the agent's internal logic, supporting operations such as halting high-risk actions or enforcing human confirmation. We argue that this Watcher paradigm holds strong potential to serve as a foundational building block for securing next-generation autonomous agent systems. Extensive qualitative and quantitative evaluations demonstrate the effectiveness and robustness of ClawKeeper across diverse threat scenarios. We release our code.

cs.CR↗

arXiv2026

Compact Kähler manifolds with partially semi-positive curvature

In this paper, we study MRC fibrations of compact Kähler manifolds with partially semi-positive curvature. We first prove that a compact Kähler manifold is rationally connected if its tangent bundle is BC-$p$ positive for all $1\leq p\leq \dim X$. As applications, we confirm a conjecture that any compact Kähler manifold with positive orthogonal Ricci curvature must be rationally connected, and generalize a result of Heier-Wong and Yang to the conformally Kähler case. The second result concern structure theorems for two immediate curvature conditions. We prove that, a compact Kähler manifold with $k$-semi-positive Ricci curvature or semi-positive $k$-scalar curvature, either the rational dimension $\geq n-k+1$ or it admits a locally constant fibration $f: X\rightarrow Y$ such that the fibre is rationally connected and the image $Y$ is Ricci-flat.

math.DG↗

arXiv2026

Non-abelian Hodge correspondence over singular Kähler spaces

In this paper, we establish the non-abelian Hodge correspondence over compact Kähler spaces with Kawamata log terminal (klt) singularities as well as over their regular loci, thereby extending the result of Greb-Kebekus-Peternell-Taji for projective klt varieties to the context of compact Kähler klt spaces. The proof relies on two key ingredients: first, we establish an equivalence over the regular loci-via harmonic bundles-between polystable Higgs bundles with vanishing orbifold Chern numbers and semi-simple flat bundles; second, we prove a descent theorem for semistable Higgs bundles with vanishing Chern classes along resolutions of singularities. As an application of our framework, we obtain a quasi-uniformization theorem for projective klt varieties with big canonical divisor that satisfy the orbifold Miyaoka-Yau equality.

math.DG↗

arXiv2026

Exploiting Double-Bounce Paths in Snapshot Radio SLAM: Bounds, Algorithms and Experiments

Radio-based simultaneous localization and mapping (SLAM) has the potential to provide precise user equipment (UE) localization and environmental sensing capabilities by exploiting radio signals. Most existing approaches leverage line-of-sight (LoS) and single-bounce non-line-of-sight (NLoS) paths solely, while higher-order NLoS paths are treated as disturbance. In this paper, we investigate the benefits of leveraging double-bounce NLoS paths for solving the bistatic snapshot radio SLAM problem. We derive the Cramer-Rao bound (CRB) for joint estimation of the UE state and landmark positions when double-bounce NLoS paths are present. In addition, we propose an algorithm to identify double-bounce NLoS paths and leverage them into joint UE and landmarks estimation. The derived bounds are validated through simulated data, and the proposed algorithms are evaluated using experimental millimeter wave (mmWave) measurements harnessing beamformed 5G cellular reference signals. The numerical and experimental results demonstrate that the double-bounce NLoS paths which share at least one incidence point (IP) with the single-bounce NLoS paths improve the estimation accuracy of the UE state and existing IPs of single-bounce NLoS paths. Importantly, exploiting double-bounce NLoS paths enhances environmental mapping capabilities by revealing landmarks that are unobservable with single-bounce NLoS paths alone.

eess.SP↗

arXiv2026

Layer-polarized Transport via Gate-defined 1D and 0D PN Junctions in Double Bilayer Graphene

We fabricate twisted double bilayer graphene devices with zero twist angle and a set of local top and bottom gates aligned perpendicularly to each other. A 1D PN junction can be electrostatically defined when the gate voltages applied to the top gates are the same but different on the bottom gates. Resistance peaks are observed at finite doping instead of at the charge neutrality points, exhibiting an unconventional broken-cross shape that arises from layer polarization of the P and N region, which can be further enhanced with finite magnetic fields. A 0D point junction (PJ) can be electrostatically defined by applying different gate voltages to the top and bottom gates, such that the P and N sides of the device are connected at a single point in the center of the device. As finite magnetic field B increases, the quantum Hall (QH) states are selectively brought into contact or away from each other depending on their layer polarization, leading to unconventional quantum oscillations which characterize the layer-polarized band-crossing. Our work provides new insights into understanding band-structure evolution and layer polarization in twisted bilayers and paves the way for new device functionality based on manipulating layer-polarized electronic states.

cond-mat.mes-hall↗

arXiv2026

Learning Hierarchical Sparse Transform Coding for 3DGS Compression

Current 3DGS compression methods largely forego the neural analysis-synthesis transform, which is a crucial component in learned signal compression systems. As a result, redundancy removal is left solely to the entropy coder, overburdening the entropy coding module and reducing rate-distortion (R-D) performance. To fix this critical omission, we propose a training-time transform coding (TTC) method that adds the analysis-synthesis transform and optimizes it jointly with the 3DGS representation and entropy model. Concretely, we adopt a hierarchical design: a channel-wise KLT for decorrelation and energy compaction, followed by a sparsity-aware neural transform that reconstructs the KLT residuals with minimal parameter and computational overhead. Experiments show that our method delivers strong R-D performance with fast decoding, offering a favorable BD-rate-decoding-time trade-off over SOTA 3DGS compressors.

cs.CV↗

arXiv2026

Mobile-Agent-v3.5: Multi-platform Fundamental GUI Agents

The paper introduces GUI-Owl-1.5, the latest native GUI agent model that features instruct/thinking variants in multiple sizes (2B/4B/8B/32B/235B) and supports a range of platforms (desktop, mobile, browser, and more) to enable cloud-edge collaboration and real-time interaction. GUI-Owl-1.5 achieves state-of-the-art results on more than 20+ GUI benchmarks on open-source models: (1) on GUI automation tasks, it obtains 56.5 on OSWorld, 71.6 on AndroidWorld, and 48.4 on WebArena; (2) on grounding tasks, it obtains 80.3 on ScreenSpotPro; (3) on tool-calling tasks, it obtains 47.6 on OSWorld-MCP, and 46.8 on MobileWorld; (4) on memory and knowledge tasks, it obtains 75.5 on GUI-Knowledge Bench. GUI-Owl-1.5 incorporates several key innovations: (1) Hybird Data Flywheel: we construct the data pipeline for UI understanding and trajectory generation based on a combination of simulated environments and cloud-based sandbox environments, in order to improve the efficiency and quality of data collection. (2) Unified Enhancement of Agent Capabilities: we use a unified thought-synthesis pipeline to enhance the model's reasoning capabilities, while placing particular emphasis on improving key agent abilities, including Tool/MCP use, memory and multi-agent adaptation; (3) Multi-platform Environment RL Scaling: We propose a new environment RL algorithm, MRPO, to address the challenges of multi-platform conflicts and the low training efficiency of long-horizon tasks. The GUI-Owl-1.5 models are open-sourced, and an online cloud-sandbox demo is available at https://github.com/X-PLUG/MobileAgent.

cs.AI↗

arXiv2026

The Devil Behind Moltbook: Anthropic Safety is Always Vanishing in Self-Evolving AI Societies

The emergence of multi-agent systems built from large language models (LLMs) offers a promising paradigm for scalable collective intelligence and self-evolution. Ideally, such systems would achieve continuous self-improvement in a fully closed loop while maintaining robust safety alignment--a combination we term the self-evolution trilemma. However, we demonstrate both theoretically and empirically that an agent society satisfying continuous self-evolution, complete isolation, and safety invariance is impossible. Drawing on an information-theoretic framework, we formalize safety as the divergence degree from anthropic value distributions. We theoretically demonstrate that isolated self-evolution induces statistical blind spots, leading to the irreversible degradation of the system's safety alignment. Empirical and qualitative results from an open-ended agent community (Moltbook) and two closed self-evolving systems reveal phenomena that align with our theoretical prediction of inevitable safety erosion. We further propose several solution directions to alleviate the identified safety concern. Our work establishes a fundamental limit on the self-evolving AI societies and shifts the discourse from symptom-driven safety patches to a principled understanding of intrinsic dynamical risks, highlighting the need for external oversight or novel safety-preserving mechanisms.

cs.CL↗

arXiv2026

Learning Grouped Lattice Vector Quantizers for Low-Bit LLM Compression

Large Language Models (LLMs) have demonstrated remarkable capabilities but typically require extensive computational resources and memory for inference. Post-training quantization (PTQ) can effectively reduce these demands by storing weights in lower bit-width formats. However, standard uniform quantization often leads to notable performance degradation, particularly in low-bit scenarios. In this work, we introduce a Grouped Lattice Vector Quantization (GLVQ) framework that assigns each group of weights a customized lattice codebook, defined by a learnable generation matrix. To address the non-differentiability of the quantization process, we adopt Babai rounding to approximate nearest-lattice-point search during training, which enables stable optimization of the generation matrices. Once trained, decoding reduces to a simple matrix-vector multiplication, yielding an efficient and practical quantization pipeline. Experiments on multiple benchmarks show that our approach achieves a better trade-off between model size and accuracy compared to existing post-training quantization baselines, highlighting its effectiveness in deploying large models under stringent resource constraints. Our source code is available on GitHub repository: https://github.com/xzhang9308/GLVQ.

cs.LG↗

arXiv2026

When Reasoning Leaks Membership: Membership Inference Attack on Black-box Large Reasoning Models

Large Reasoning Models (LRMs) have rapidly gained prominence for their strong performance in solving complex tasks. Many modern black-box LRMs expose the intermediate reasoning traces through APIs to improve transparency (e.g., Gemini-2.5 and Claude-sonnet). Despite their benefits, we find that these traces can leak membership signals, creating a new privacy threat even without access to token logits used in prior attacks. In this work, we initiate the first systematic exploration of Membership Inference Attacks (MIAs) on black-box LRMs. Our preliminary analysis shows that LRMs produce confident, recall-like reasoning traces on familiar training member samples but more hesitant, inference-like reasoning traces on non-members. The representations of these traces are continuously distributed in the semantic latent space, spanning from familiar to unfamiliar samples. Building on this observation, we propose BlackSpectrum, the first membership inference attack framework targeting the black-box LRMs. The key idea is to construct a recall-inference axis in the semantic latent space, based on representations derived from the exposed traces. By locating where a query sample falls along this axis, the attacker can obtain a membership score and predict how likely it is to be a member of the training data. Additionally, to address the limitations of outdated datasets unsuited to modern LRMs, we provide two new datasets to support future research, arXivReasoning and BookReasoning. Empirically, exposing reasoning traces significantly increases the vulnerability of LRMs to membership inference attacks, leading to large gains in attack performance. Our findings highlight the need for LRM companies to balance transparency in intermediate reasoning traces with privacy preservation.

cs.CR↗

arXiv2026

From static to adaptive: immune memory-based jailbreak detection for large language models

Large Language Models (LLMs) serve as the backbone of modern AI systems, yet they remain susceptible to adversarial jailbreak attacks. Consequently, robust detection of such malicious inputs is paramount for ensuring model safety. Traditional detection methods typically rely on external models trained on fixed, large-scale datasets, which often incur significant computational overhead. While recent methods shift toward leveraging internal safety signals of models to enable more lightweight and efficient detection. However, these methods remain inherently static and struggle to adapt to the evolving nature of jailbreak attacks. Drawing inspiration from the biological immune mechanism, we introduce the Immune Memory Adaptive Guard (IMAG) framework. By distilling and encoding safety patterns into a persistent, evolvable memory bank, IMAG enables adaptive generalization to emerging threats. Specifically, the framework orchestrates three synergistic components: Immune Detection, which employs retrieval for high-efficiency interception of known jailbreak attacks; Active Immunity, which performs proactive behavioral simulation to resolve ambiguous unknown queries; Memory Updating, which integrates validated attack patterns back into the memory bank. This closed-loop architecture transitions LLM defense from rigid filtering to autonomous adaptive mitigation. Extensive evaluations across five representative open-source LLMs demonstrate that our method surpasses state-of-the-art (SOTA) baselines, achieving a superior average detection accuracy of 94\% across diverse and complex attack types.

cs.CR↗

arXiv2026

SecMoE: Communication-Efficient Secure MoE Inference via Select-Then-Compute

Privacy-preserving Transformer inference has gained attention due to the potential leakage of private information. Despite recent progress, existing frameworks still fall short of practical model scales, with gaps up to a hundredfold. A possible way to close this gap is the Mixture of Experts (MoE) architecture, which has emerged as a promising technique to scale up model capacity with minimal overhead. However, given that the current secure two-party (2-PC) protocols allow the server to homomorphically compute the FFN layer with its plaintext model weight, under the MoE setting, this could reveal which expert is activated to the server, exposing token-level privacy about the client's input. While naively evaluating all the experts before selection could protect privacy, it nullifies MoE sparsity and incurs the heavy computational overhead that sparse MoE seeks to avoid. To address the privacy and efficiency limitations above, we propose a 2-PC privacy-preserving inference framework, \SecMoE. Unifying per-entry circuits in both the MoE layer and piecewise polynomial functions, \SecMoE obliviously selects the extracted parameters from circuits and only computes one encrypted entry, which we refer to as Select-Then-Compute. This makes the model for private inference scale to 63$\times$ larger while only having a 15.2$\times$ increase in end-to-end runtime. Extensive experiments show that, under 5 expert settings, \SecMoE lowers the end-to-end private inference communication by 1.8$\sim$7.1$\times$ and achieves 1.3$\sim$3.8$\times$ speedup compared to the state-of-the-art (SOTA) protocols.

cs.CR↗