Searcharxiv⌕ Search

arXiv subjects

Mingzhe Li

Publications and source records attributed to Mingzhe Li.

At least 19 recordsLinked to original sources
arXiv2026

MS-COOT: Comparing Morse-Smale Complexes with Co-Optimal Transport

Understanding and comparing structures in scalar fields is a central challenge in scientific visualization, with applications ranging from feature analysis to temporal and structural comparison. The Morse-Smale (MS) complex provides a natural representation by decomposing a scalar field into regions induced by gradient flow. However, existing approaches typically rely on graph-based representations, matching critical points without explicitly establishing correspondences between regions. In this work, we represent the MS complex as a hypergraph, where critical points form nodes and regions define hyperedges. We introduce MS-COOT, a co-optimal transport distance that jointly computes correspondences between critical points and regions. This formulation enables explicit region-to-region matching within a distance-based framework, allowing identification of region-level events such as splitting and merging. We instantiate this framework with domain-specific components, including a hypernetwork function encoding relationships between critical points and regions, persistence-based probability measures, and a sample cost term that incorporates critical point attributes. We evaluate MS-COOT on datasets spanning 2D simulations, 3D surface meshes, and volumetric data. Our results show that MS-COOT provides distinct characterization of certain region-level structural changes compared to graph-based distances, achieves competitive accuracy in shape classification, and improves classification of simulation parameter settings in a 3D ensemble.

cs.GR↗
arXiv2026

Source or It Didn't Happen: A Multi-Agent Framework for Citation Hallucination Detection

Large language models are increasingly used in scientific writing, yet they can fabricate citation-shaped references that appear plausible but fail bibliographic verification. Existing detectors often reduce verification to binary found/not-found decisions and rely on brittle parsing or incomplete retrieval, offering little field-level signal to auditors. We reframe citation hallucination detection as taxonomy-aligned field-level adjudication and introduce a 12-code taxonomy spanning Real, Potential, and Hallucinated citations. Based on this taxonomy, we build CiteTracer, a cascading multi-agent detector that extracts structured citations from PDF and BibTeX, retrieves evidence through cache lookup, URL fetch, scholar connectors, and web search, applies deterministic field matching, and routes ambiguous cases to class-specialist judgers. We release a benchmark of 2,450 synthetic citations built from real seeds with controlled LLM mutations, paired with 957 real-world fabricated citations drawn from desk-rejected submissions at ICLR 2026 and an anonymous conference. CiteTracer reaches 97.1% accuracy on the synthetic benchmark, with class-level F1 scores of 97.0, 95.8, and 98.5 for Real, Potential, and Hallucinated, respectively, and detects 97.1% of fabrications on the real-world set without abstaining. Code: https://github.com/aaFrostnova/CiteTracer.

cs.CL↗
arXiv2026

Offline Policy Optimization with Posterior Sampling

A fundamental challenge in model-based offline reinforcement learning (RL) lies in the trade-off between generalization and robustness against exploitation errors in out-of-distribution (OOD) regions. The key to resolving this trade-off lies in enabling the model to explore OOD regions that remain consistent with underlying physical dynamics. However, achieving this is challenging because limited data cannot uniquely identify the dynamics model, and unconstrained exploration is risky. Existing methods often overlook this nuance, addressing the risk through excessive pessimistic regularization, which ensures robustness but sacrifices generalization. To address this, we propose PSPO, which treats the dynamics model as a random variable rather than a point estimate. This formulation inherently allows for controlled exploration of OOD regions. By alternately updating the posterior distribution and the policy, we design a regularized optimization algorithm with convergence guarantees. Experiments on standard benchmarks demonstrate that PSPO achieves superior performance compared to state-of-the-art baselines. Further analysis confirms that our method attains the desired pessimism-free property while maintaining robustness, and ablation studies verify the effectiveness of each proposed module.

cs.AI↗
arXiv2026

Made to Measure: Designing Image Watermarks to Specification

Image watermarking supports provenance and attribution by embedding verifiable identity information into images. Practical deployments, however, must jointly satisfy requirements for attack resistance, false-positive rate (FPR), image quality, and latency. Existing watermarking methods are robust to different classes of transformations, so combining complementary methods can provide broader protection than any single watermark. Such composition is challenging, as additional fragments increase distortion and decoding cost and must share the same FPR budget. Therefore, we propose **TAILOR**, a request-conditioned watermark composition framework with three stages: (1) *offline characterization* measures fragment recovery, distortion, and runtime as response curves over embedding strength; (2) *joint configuration selection* encodes the request as an SMT model over these curves and solves for the lowest-distortion composition of fragments, strengths, order, and geometric recovery; and (3) *live calibration* validates the selected configuration on the user's images and refines predictions that fail to transfer. Experimental results across 7,321 distinct requests spanning five scenarios and 20 attack settings show that **TAILOR** achieves **96.21%** scenario-averaged request satisfaction with a mean PSNR of **41.02 dB**, outperforming existing methods in robustness while achieving consistently better image quality. Code is available at [https://github.com/aaFrostnova/Tailor](https://github.com/aaFrostnova/Tailor).

cs.CR↗
arXiv2026

LITERARYBIGFIVE: Author-Personalized Text Generation in a Unified Interpretable Space

Personalized text generation for authors and literary writing is essential for applications such as adaptive writing assistants, creative support tools, and computational literary analysis. However, existing approaches to author modeling and personalization often represent writing behavior as independent labels, requiring large-scale corpus collection or fine-tuning for each author or stylistic category. Such formulations are costly, difficult to interpret, and poorly suited for generalizing across authors. Inspired by the Big Five model's dimensional view of personality, we propose LiteraryBigFive, a framework that reframes authorial writing characteristics as coordinates within a unified and interpretable space. In this space, we derive each interpretable axis (e.g., Classicism, Emotionality) from activation-space contrasts between author-written and neutral passages, yielding distinct stylistic dimensions that allow texts or authors to be positioned within a five-dimensional system. Beyond localizing different authors, we further introduce an interpretable steering mechanism, which adaptively guides text generation toward target coordinates to perform author-personalized writing. Experimental results show that LiteraryBigFive improves authorial expressiveness while preserving semantic fidelity. The derived author per-axis scores strongly correlate with real-world literary consensus, offering transparent and interpretable explanations of author-specific generation behavior: https://github.com/Znull-1220/LiteraryBigFive.

cs.CL↗
arXiv2026

Disappearing Ink: Obfuscation Breaks N-gram Code Watermarks in Theory and Practice

Large language models (LLMs) are increasingly used for code generation, making reliable identification of machine-generated code important for attribution, tracking, and misuse detection. Existing code watermarking methods are dominated by N-gram-based schemes, yet their robustness has mostly been evaluated only against simple edits or optimizations. We argue that this significantly overstates security, because software engineering already provides stronger semantics-preserving transformations in the form of code obfuscation. We study N-gram-based code watermarking under obfuscation. We formally model semantics-preserving transformations as a Markov random walk and prove that, under an intuitive and experimentally supported assumption called distribution consistency, obfuscation can nullify the robustness of N-gram-based watermarks. If the original detector has a false positive rate fpr, then after obfuscation, its failure rate on watermarked code approaches 1 - fpr. We validate this theory on three state-of-the-art watermarking schemes, two LLMs, two programming languages, four benchmarks, and four obfuscators. Across all settings, detectors collapse to near-random performance on obfuscated code (AUROC tightly around 0.5), and for each language, at least one attack leaves all post-obfuscation AUROC scores below 0.6. These results jointly show that current N-gram-based code watermarks are not robust to realistic obfuscation attacks and motivate more semantics-aware alternatives.

cs.CR↗
arXiv2026

Preserve, Then Resolve: Many-to-Many Association and Robust Estimation with General-Purpose Visual Features

The semantic transferability of general-purpose visual features does not guarantee geometric consistency across images. Using frozen DINOv3 features, we show that geometrically correct correspondences often fall below rank one in cosine similarity yet remain within a small top-$K$ candidate set. This motivates a preserve-then-resolve design: we retain multiple candidates in a many-to-many (m-to-m) association graph and defer their disambiguation to robust estimation. We study m-to-m robust estimation from a probabilistic perspective. We interpret the existing Matching Cardinality Maximization (MCM) mechanism as a dominant-cardinality approximation to likelihood maximization and propose a faster, real-valued mechanism called Harmonic Consensus Maximization (HCM). A two-stage LO-RANSAC uses HCM for candidate sourcing and MCM for graph-aware selection. We evaluate end-task gains in relative-pose estimation, where our pipeline consistently improves rank-one baselines for DINOv2, DINOv3, V-JEPA 2.1, and SigLIP 2. Code is available at https://github.com/LIAS-CUHKSZ/preserve_then_resolve.

cs.CV↗
arXiv2026

The effects on CMB power spectra and bispectra from the polarization rotation and its correlations with temperature and E-polarization

The Chern-Simons term, through which the cosmic Axion-like field couples to the electromagnetic field, has the effect to rotate CMB polarization directions and to break the CPT symmetry. This rotation will change the CMB power spectra, no matter isotropic or anisotropic the rotation angle is. In this paper we revisit this issue by further considering the correlations between the (anisotropic) rotation angle $α$ and the CMB temperature and (unrotated) $E$ polarization fields. These correlations could be generated in the Axion-like models with nonzero potential under the adiabatic initial condition. We first investigate how these correlations contribute further modifications to the CMB power spectra, then calculate the CMB bispectra for the temperature and rotated polarization fields. These bispectra would vanish if the $Tα$ and $Eα$ correlations are absent. So, they are useful in searching for CPT violation and the $Tα$ and $Eα$ correlations arisen in the Axion-like models.

astro-ph.CO↗
arXiv2026

Study the nature of dynamical dark energy by measuring the CMB polarization rotation angle

Recent results from the Dark Energy Spectroscopic Instrument (DESI) support the dynamical dark energy. Intriguingly, the data favor a transition of the dark energy equation of state across $w=-1$, a hallmark of the Quintom scenario. In this paper, we consider a different approach to the dynamical nature of dark energy by investigating its interaction with ordinary matters, specifically the Chern-Simons (CS) interaction with photons. In cosmology, this interaction rotates the polarized plane of the cosmic microwave background (CMB) photons, which induces non-zero polarized TB and EB power spectra. We forecast this measurement with the Ali CMB Polarization Telescope (AliCPT) experiment. We take the best-fit value of the isotropic rotation angle from Planck data as our fiducial input. We project that 11 module-year (modyr) of observations will yield an improved detection sensitivity with a significance $\sim 5σ$, given a calibration precision of $0.1^\circ$ in the polarization angle. We also forecast AliCPT's sensitivity to the amplitude of a scale invariant spectrum of the anisotropic polarization rotation field. With $50$~modyr of observations, the large-aperture configuration is expected to reach $σ_{A_{\mathrm{CB}}}\sim10^{-2}$, offering a sixfold improvement over the small-aperture design and enabling competitive tests of spatial fluctuations in the dark energy field.

astro-ph.CO↗
arXiv2026

AI Can Learn Scientific Taste

Scientific discovery depends on expert judgement and foresight, which we call scientific taste: the ability to judge and propose research ideas with the potential for long-term scientific impact. Scientific taste is largely concentrated among highly experienced researchers, whose expertise is usually limited to a few specialised fields. If AI could learn scientific taste, it could reduce reliance on human experts and accelerate scientific discovery. Whether AI can learn this ability remains an open question. We introduce Reinforcement Learning from Community Feedback (RLCF) to learn judgement and ideation. Scientific Judge learns from community feedback, such as citations. Scientific Thinker learns to propose research ideas with high potential impact. Experiments show that Scientific Judge outperforms strong LLM baselines and that learned judgement generalises to future-year papers, other community metrics, and unseen fields. Furthermore, Scientific Thinker proposes research ideas with higher potential impact than those proposed by baselines. These results suggest that AI can learn scientific taste, marking an important step towards AI systems that could help accelerate scientific discovery.

cs.CL↗
arXiv2026

MetaState: Persistent Working Memory Enhances Reasoning in Discrete Diffusion Language Models

Discrete diffusion language models (dLLMs) generate text by iteratively denoising a masked sequence. However, standard dLLMs condition each denoising step solely on the current hard-masked sequence, while intermediate continuous representations are discarded after sampling and remasking. We term this bottleneck the \textbf{Information Island} issue: continuous information remains isolated within individual denoising steps and fails to propagate across the trajectory. This bottleneck is especially harmful for reasoning, which requires intermediate reasoning state to be preserved and updated across many denoising steps. To address this limitation, we introduce \textbf{MetaState}, a lightweight recurrent augmentation that equips a frozen dLLM backbone with persistent, fixed-size working memory. MetaState comprises three modules with a shared time conditioner: a cross-attention \textbf{Mixer} that reads backbone activations into memory slots, a GRU-style \textbf{Updater} that integrates information across steps, and a cross-attention \textbf{Injector} that writes the updated memory back into the backbone. We train these modules with a dedicated $K$-step unrolling pipeline to learn multi-step dynamics. MetaState adds only ${\sim}0.6\%$ trainable parameters while keeping the backbone frozen, and consistently improves reasoning performance over frozen baselines on mathematical reasoning and code generation benchmarks, with an average gain of 4.5 percentage points across all evaluations.

cs.CL↗
arXiv2026

One Model, Multiple Goals: Adaptive Multi-Objective Learning for E-commerce Dialogue Systems

Dialogue systems in e-commerce scenarios often need to satisfy multiple objectives: accurately reasoning over user profiles (e.g., eligibility, credit limit) to ensure correct decision-making and user state interpretation, while also generating natural and faithful responses. These goals are complementary but not identical. In this work, we propose MORE, an adaptive Multi-Objective REinforcement learning framework that jointly optimizes reasoning accuracy and linguistic naturalness. Our preliminary experiments show that directly mixing rewards with diverging optimization dynamics can cause oscillations and unstable learning. Thus, instead of optimizing a single mixed reward, we treat reasoning functions as constraints that guide policy optimization. At inference time, the system directly generates responses without explicit reasoning steps, while still benefiting from reasoning-enhanced scaffold and avoiding additional inference overhead. To better balance linguistic objectives during response generation, we introduce an adaptive multi-reward mechanism that aggregates signals such as fluency and naturalness and dynamically reweighs them via gradient feedback. We evaluate MORE on two real-world dialogue systems at ByteDance and the MultiWOZ 2.2 benchmark, where it consistently outperforms strong baselines. In 14-day online experiments on ByteDance production traffic, MORE improves overall and reached conversion by 16.53% and 30.09%, while increasing user satisfaction and reducing handoff rates. Notably, in a human-machine comparison, MORE recovers about 60% of the incremental conversion lift achieved by human agents.

cs.CL↗
arXiv2026

Realization of quintom dark energy after DESI DR2 in Nieh-Yan modified teleparallel gravity

Recent observations from the DESI Collaboration indicate a preference for quintom dark energy, i.e., its equation of state evolves across the cosmological constant boundary $w=-1$. It is well known that models with single perfect fluid or single scalar field minimally coupled to Einstein gravity develop perturbative instabilities around the crossing, thereby cannot realize the quintom scenario. In this paper, we propose a method to circumvent the instability problem of these models by considering the coupling of dark energy to the Nieh-Yan density within the framework of teleparallel gravity. We show that with this coupling the background evolution is not affected, but the dark energy perturbation is removed from the menu of dynamical degrees of freedom, thus avoiding the inherent difficulties in the old models. Furthermore, the Nieh-Yan coupling causes parity violation in gravitational waves, and this can be considered as a clear prediction of this mechanism.

gr-qc↗
arXiv2026

Membership Inference Attacks on Vision-Language-Action Models

Membership inference attacks (MIAs) have been extensively studied in large language models (LLMs) and vision-language models (VLMs), yet their implications for vision-language-action (VLA) models remain largely unexplored. VLA models differ from standard LLMs and VLMs in several important ways: they are often fine-tuned for many epochs on relatively small embodied datasets, operate over constrained and structured action spaces, and expose action outputs that can be observed as executable behaviors and temporally correlated trajectories. These characteristics suggest a distinct and potentially more informative attack surface for membership inference. In this work, we present the first systematic study of MIAs against VLA systems. We formalize two membership inference settings for VLA models: sample-level inference over individual transition samples and trajectory-level inference over complete embodied demonstrations. We further develop a suite of attack methods under multiple access regimes, including strict black-box access. Our attacks exploit both classic MIA signals, such as token likelihood, and VLA-specific signals, such as observable action errors and temporal motion patterns. Across multiple VLA benchmarks and representative VLA models, these attacks achieve strong inference performance, showing that VLA models are highly vulnerable to membership inference. Notably, black-box attacks based only on generated actions achieve strong performance, highlighting a practical privacy risk for deployed embodied AI systems. Our findings reveal a previously underexplored privacy risk in robotic and embodied AI, and underscore the need for dedicated privacy evaluation and defenses for VLA models.

cs.CR↗
arXiv2026

When Personalization Tricks Detectors: The Feature-Inversion Trap in Machine-Generated Text Detection

Large language models (LLMs) have grown more powerful in language generation, producing fluent text and even imitating personal style. Yet, this ability also heightens the risk of identity impersonation. To the best of our knowledge, no prior work has examined personalized machine-generated text (MGT) detection. In this paper, we introduce \dataset, the first benchmark for evaluating detector robustness in personalized settings, built from literary and blog texts paired with their LLM-generated imitations. Our experimental results demonstrate large performance gaps across detectors in personalized settings: some state-of-the-art models suffer significant drops. We attribute this limitation to the \textit{feature-inversion trap}, where features that are discriminative in general domains become inverted and misleading when applied to personalized text. Based on this finding, we propose \method, a simple and reliable way to predict detector performance changes in personalized settings. \method identifies latent directions corresponding to inverted features and constructs probe datasets that differ primarily along these features to evaluate detector dependence. Our experiments show that \method can accurately predict both the direction and the magnitude of post-transfer changes, showing 85\% correlation with the actual performance gaps. We hope that this work will encourage further research on personalized text detection.

cs.CL↗
arXiv2026

The Quintom theory of dark energy after DESI DR2

Observations from DESI DR2 are challenging the $Λ$CDM paradigm by suggesting that the equation-of-state parameter of dark energy evolves across $w = -1$, a phenomenon known as the Quintom scenario. Inspired by this development, we present a staged review of Quintom cosmology including its theoretical foundations, observational supports, and implications as well as possible extensions. We first trace the historical progression from Einstein's static cosmological constant to modern dynamical dark energy, summarizing recent cosmological constraints that favor an evolving $w(z)$ along time. A key focus is the theoretical no-go theorem for dark energy showing that no single canonical field or perfect fluid model can smoothly cross the $w = -1$ boundary. We then survey viable Quintom constructions, including two-field models, single-scalar fields with higher derivatives, modified gravity frameworks, interacting dark energy, and an effective field theory approach that unifies these mechanisms. Possible interactions of Quintom fields with ordinary matter and the potential roles in yielding non-singular universe solutions are discussed.

astro-ph.CO↗
arXiv2026

Thinking with Video: Video Generation as a Promising Multimodal Reasoning Paradigm

The "Thinking with Text" and "Thinking with Images" paradigms significantly improve the reasoning abilities of large language models (LLMs) and Vision-Language Models (VLMs). However, these paradigms have inherent limitations. (1) Images capture only single moments and fail to represent dynamic processes or continuous changes, and (2) The separation of text and vision as distinct modalities, which hinders unified multimodal understanding and generation. Therefore, we propose "Thinking with Video", a new paradigm that leverages video generation models such as Sora-2 to use video frames as a unified medium for multimodal reasoning. To support this exploration, we developed the Video Thinking Benchmark (VideoThinkBench), which covers both vision-centric tasks (e.g., Eyeballing Puzzles) and text-centric tasks (e.g., GSM8K and MMMU). Our evaluation on VideoThinkBench establishes Sora-2 as a capable reasoner. On vision-centric tasks, Sora-2 is comparable to state-of-the-art (SOTA) VLMs, and even surpasses GPT-5 by 10% on eyeballing puzzles. On text-centric tasks, Sora-2 achieves 92% accuracy on MATH, and 69.2% accuracy on MMMU. Furthermore, we systematically analyze the source of these abilities. We also find that self-consistency and in-context learning can improve Sora-2's performance. In summary, our findings show that the video generation model is the potential unified multimodal understanding and generation model, positioning "Thinking with Video" as a potential unified multimodal reasoning paradigm.

cs.CV↗
arXiv2026

Magnetic-field-tunable commensurate multi-q charge orders on UTe2 (011) surface

The heavy-fermion superconductor UTe2 has attracted intense interest as a candidate for spin-triplet pairing. Recent scanning tunneling microscopy (STM) studies have reported complex charge orders (COs) on its (011) surface, but their origin and relationship with superconductivity remain controversial. Here, by performing temperature-, magnetic field-, and sample-dependent STM measurements, we identify multiple new CO wave vectors beyond those previously reported. All these CO wave vectors are strictly locked to integer multiples of 1/14 and 1/4 of the reciprocal lattice vectors of the UTe2 (011) surface, and multiple of them coexist in real space, collectively revealing a family of field-tunable, commensurate multi-q COs. These COs exist within an energy range much larger than the superconducting energy scale, their emergence suppresses the density of states near EF, yet show negligible coupling to bulk superconductivity and magnetic vortices. Our findings strongly disfavor the Fermi surface nesting or primary pair-density-wave pictures, but are consistent with a surface parent spin order.

cond-mat.supr-con↗