arXiv subjects

Li Zhang

Publications and source records attributed to Li Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Estimating redshift distributions of Fermi blazars with multiple machine-learning models

Approximately half of the Fermi-LAT-detected blazars lack spectroscopic redshift measurements, which limits population studies and investigations of the cosmological evolution of the gamma-ray blazar population. We aim to develop a multi-model ensemble framework to systematically evaluate the performance of machine-learning models for blazar redshift estimation and to provide predictive redshift probability distributions with quantified uncertainties for all blazars without measured redshifts. We construct an ensemble of 35 machine-learning regression models in the $1/(1+z)$ space. Predictions from individual models are combined to build non-parametric redshift probability density functions (PDFs) for each source, allowing an explicit characterization of predictive uncertainty. The source-level PDFs are subsequently stacked to derive population-level redshift distributions. Across the diverse models within our framework, we find comparable predictive performance, indicating a feature-limited performance plateau. We compile a catalog of predictive redshift distributions for 1,590 blazars without measured redshifts. The stacked PDFs recover the characteristic double-peaked redshift structure of blazars without using source class labels. For blazar candidates of uncertain type, the predicted redshift distribution exhibits a bimodal structure, suggesting a slightly larger contribution from FSRQ-like sources. Our multi-model statistical framework achieves competitive predictive performance while retaining faint sources. We provide predicted redshift distributions with quantified uncertainties for 1,590 blazars without measured redshifts, offering improved support for population and evolutionary studies. More broadly, this approach highlights the practical value of uncertainty-aware multi-model regression ensembling in astronomical data analysis.

astro-ph.HE↗

arXiv2026

Brain-Token Learning: Microstate-Based Tokenization and Multi-Scale Interaction for Long-Horizon EEG Sequence Modeling

Electroencephalography (EEG) provides a non-invasive window into dynamic brain activity, yet modeling long-horizon EEG sequences remains challenging due to their high temporal complexity, substantial variability across subjects, and the lack of biologically meaningful sequence representations. Existing tokenization strategies, such as fixed-window and patch-based representations, discretize EEG signals according to artificial temporal boundaries, which may disrupt intrinsic brain-state dynamics. In this work, we propose Brain-Token Learning, a neuroscience-inspired framework that introduces Brain Tokenization for long-horizon EEG sequence modeling. Instead of partitioning EEG signals into predefined temporal segments, Brain Tokenization represents EEG as sequences of recurrent microstate-derived brain tokens, where each token corresponds to a quasi-stable large-scale brain state with variable temporal duration. Based on these biologically grounded tokens, we further develop a multi-scale token interaction module consisting of Latent State Aggregation and State Transition Modeling to jointly capture global brain-state context and local microstate transitions. We evaluate Brain-Token on five heterogeneous EEG datasets, including the newly collected long-horizon NeuroLong dataset and four affective or clinical EEG datasets (SEED, DEAP, MDD, and NSSI). Extensive experiments demonstrate that Brain-Token consistently outperforms conventional CNN/LSTM architectures, Transformer-based models, and domain adaptation methods across diverse EEG scenarios. Further analysis verifies the effectiveness of microstate-based tokenization and multi-scale interaction for learning robust and interpretable EEG representations. These results establish Brain-Token as a biologically grounded tokenization paradigm for long-horizon EEG sequence modeling.

cs.AI↗

arXiv2026

PhysReflect: Geometry and Perception Guided Diffusion for Physically-Plausible Mirror Reflections

Diffusion models generate high-quality images, yet often violate the physical laws governing mirror reflections. Reflections often suffer from geometric aberrations, including positional offsets, directional misalignment, proportional imbalance, and structural distortion. These failures remain evident even in contemporary state-of-the-art generative systems. Existing methods itigate this problem through synthetic data scaling or auxiliary depth conditioning, yet their merely reliance on latent-space noise reconstruction losses as implicit supervision prevents direct enforcement of reflection-specific geometric and perceptual constraints. To bridge this gap, we present PhysReflect, a geometry and perception guided diffusion framework that decodes the predicted clean latent into pixel space at each training step and applies annealed supervision through two complementary differentiable objectives. The Geometric Loss enforces mirror-induced spatial consistency through sparse epipolar correspondence and dense boundary projection alignment, where a SAM2-based TwinTrack mechanism provides stable in-mirror localization for boundary-aware supervision. The Perceptual Loss preserves reflected appearance by combining Semantic Consistency Loss, which maintains reflected identity and appearance via DINOv2 features, and Lighting Consistency Loss, which regularizes depth, surface-normal, and illumination coherence under monocular geometry priors. Experiments on synthetic and real-world benchmarks show that PhysReflect outperforms prior mirror-reflection methods in geometric, perceptual, and physical-plausibility metrics, as well as qualitative visual results.

cs.CV↗

arXiv2026

When the Agent Becomes the Kernel: A Systematization of Security on the Path to AI-Native Operating Systems

Large language model agents are now privileged principals that take consequential actions: editing code repositories, operating inboxes, completing purchases. Their authority is kernel-grade, but it comes without what classical systems security requires: a trusted mediator interposed on every access. Operating-system vendors are now rebuilding the platform around this de-facto agent kernel, inheriting complete mediation as a design problem. We systematize the security of such systems around a single distinction: a crossing mediated over provenance admits a deterministic check, while one over content semantics does not. A trust-boundary taxonomy locates where mediation must occur and isolates the central mediation gap at two kinds of semantic judgment: distinguishing data from instruction in untrusted input, and an authorized action from an unauthorized one. We argue that this gap leaves an irreducible residual of undetected attacks wherever inputs and actions are not restricted in advance to an enumerated set. The same distinction makes attack-success statistics actionable, placing each number on a spectrum from deployment debt (a sound deterministic mediator left unused) to a structural gap (no such mediator known). We systematize defenses across runtime monitoring, architectural separation, and authorization, and show that current evaluations tend to overstate deployed security through evaluation-validity failures. Finally, we carry that analysis forward beyond the de-facto kernel, to an architecture in which the model itself becomes the arbitration core, and derive the design constraints, open challenges, and research agenda for a security-first AI-native OS.

cs.CR↗

arXiv2026

UltraTex: Unleashing 2K Multi-View Diffusion for 3D Texturing

High-quality texture generation is essential for creating realistic and production-ready 3D assets. Recent multi-view diffusion methods have shown promising results for image-guided 3D texturing, but they are typically constrained to low operating resolutions such as 512 or 768, making it difficult to preserve high-frequency details from high-resolution reference images. Scaling this paradigm to 2048 resolution is computationally prohibitive, as the unified multi-view sequence exceeds 212K tokens and incurs excessive memory and latency. In this paper, we present UltraTex, an efficient end-to-end framework for high-resolution multi-view diffusion-based 3D texturing. Our key observation is that object-centric multi-view renderings contain two major sources of redundancy: background-induced sequence redundancy and sparse token interactions within the foreground. To address them, we introduce Background Token Dropping, which removes background tokens before the DiT backbone, and Block-Sparse Attention, which reduces attention computation over the retained foreground sequence. To enable efficient foreground-only inference while avoiding reconstruction artifacts, we further design Foreground-Aware VAE Decoding to ensure the quality of the final high-resolution views. To satisfy the demanding data requirements of 2K-resolution multi-view diffusion training, we construct G-buffer TexVerse, a large-scale, ultra-high-resolution multi-view rendering dataset covering over 268,000 3D assets. Extensive experiments show that UltraTex generates visually faithful textures with rich fine-grained details, while substantially improving efficiency, achieving $20.6\times$--$91.1\times$ training speedup and $22.3\times$--$74.6\times$ end-to-end inference speedup over the baseline on common samples in our dataset. Code and data is at https://yiboz2001.github.io/UltraTex.

cs.CV↗

arXiv2026

M-SQE: Multilingual Skill Quality Estimation for Enhancing Language Equality in Agentic Skill Use

Agent skills, reusable procedural documents that extend LLM agents beyond their parametric memory, have become an important interface for deploying agents on real-world tasks. Community-maintained skill libraries built around this interface are growing rapidly. However, this ecosystem remains deeply English-centric: our audit finds that low-resource languages such as Swahili and Hindi have no in-language skill content, so retrieval often returns a skill written in a different language than the query, degrading accuracy and recall. A practical solution is to synthesize in-language skills for retrieval but the quality can be unreliable, so relevance in this setting alone often surfaces a related but unusable candidate. To address this, we propose M-SQE, a post-retrieval Multilingual Skill Quality Estimation framework that scores candidates via a Theory view for intrinsic quality and an Action view for task-grounded utility, unified into a domain-conditioned final score. We evaluate M-SQE across three skill-use domains: general, tool-use, and cultural tasks. Empirically, we build three-layer candidate skill pools mirroring today's ecosystem, where M-SQE's task success exceeds existing baseline's average by at least +3.5 points across three different retrievers. Particularly, M-SQE lifts the lowest-resource languages most (+12.9pp on Hindi and +5.6pp on Swahili) and achieves strong performance across all six culture regions, thereby moving agentic skill use toward linguistic and cultural equality.

cs.CL↗

arXiv2026

X-WBC: A Cross-Embodiment Foundation Model for Humanoid Whole-Body Control

Scaling humanoid whole-body control toward general-purpose deployment requires large human motion corpora and training experience shared across robot bodies. Existing methods usually train one policy per robot, leaving motion experience isolated across embodiments. We introduce X-WBC, a cross-embodiment foundation framework that separates relatively shared human motion semantics from embodiment-specific physical execution. Human-centered command tokens align full human motion, robot reference motion, and sparse VR observations. A causal Transformer learns reusable temporal structure from mixed multi-robot rollouts, while lightweight robot-specific modules map the shared representation to each robot's proprioception and action space. Across nine simulated embodiments, external motions, and four real robots, experiments show that joint training improves tracking, the aligned representation supports consistent control across command sources, and the learned policy remains competitive beyond the training corpus. These results support heterogeneous humanoids as joint data sources and establish cross-embodiment joint training as a practical route toward whole-body control foundation models.

cs.RO↗

arXiv2026

On the uniform flatness of polynomial graphs

This paper grew out of an effort to understand the so-called uniform flatness, an important concept in the Bargmann--Fock interpolation theory developed by Varolin et al. We show that the graph of every polynomial in one complex variable is a uniformly flat algebraic curve in $\mathbb C^2$, with vanishing upper density with respect to the Gaussian weight. Combined with a result of Pingali and Varolin, this implies that such graphs are interpolating for the Bargmann--Fock space on $\mathbb C^2$.

math.CV↗

arXiv2026

Square-shaping of sturdy optical vortex droplets in quasi-phase-matched photonic crystals

We elaborate a scheme for controllable shaping of self-trapped vortex states in a quasi-phase-matched three-dimensional photonic crystal with the combination of self-focusing quadratic and defocusing cubic material nonlinearities. The setting gives rise to sturdy droplet-like vortex modes, capable to adapt to externally imposed strong geometric constraints. The application of a square-shaped modulation in the transverse $\left(x,y\right) $ plane and periodic quasi-phase-matching to the quadratic nonlinear coefficient $d_{z}$ along the propagation direction $z$ leads to the formation of square vortex droplets (VDs) with fourfold rotational symmetry ($C_{4}$). These states preserve the vortical phase circulation and exhibit robust propagation in a broad parameter region. In the oversaturated regime dominated by the cubic self-defocusing, the square-shaped VDs obey the anti-Vakhitov--Kolokolov stability criterion. The results, which are produced, chiefly, for the VDs with topological charge $S=1$, and also, in a partial form, for $S=2$ and $4$, reveal an unexpected universality: the vortex robustness is not contingent upon the circular symmetry. Thus, the combination of the competing nonlinearities and geometric confinement provides not only an effective method for the formation of self-trapped vortex states, but also new insight into generality of the topological protection in nonlinear optical fields.

physics.optics↗

arXiv2026

Benchmarking Requirement-to-Architecture Generation with Hybrid Evaluation

Software architecture serves as the blueprint of a software system, capturing high-level structural decisions that shape downstream implementation and system quality. Despite this central role, generating architecture designs from requirement documents remains underexplored, with limited task-specific benchmarks for generation or rigorous evaluation. To bridge this gap, we introduce R2ABench, a benchmark for requirements-to-architecture (R2A) reasoning. R2ABench contains 68 projects with human-validated references collected from both educational-style settings and GitHub repositories. Each project is packaged as a unified instance that provides a structured software requirements specification (SRS), a reference architecture view in PlantUML, and the corresponding source requirement artifacts. To evaluate generated architecture views, we design a layered hybrid evaluation framework spanning syntax validation, structural graph diagnostics, and semantic and evidence-based architecture scoring. Using this framework, we conduct a comprehensive empirical study of state-of-the-art LLMs and agents on R2ABench. We find that current systems can often generate syntactically valid and readable architecture views, but still struggle with relation-level architecture modeling across both subsets: component identification is substantially stronger than edge recovery, and edge hallucination is the dominant structural failure mode. Semantically, structural fidelity does not guarantee requirement coverage or traceability, which emerge as the primary quality gaps. The associated data files are available at https://figshare.com/s/01f0a5fb6243a6a60f23.

cs.SE↗

arXiv2026

Who You Are Adds Nothing Detectable to Where You Go Next: Sociodemographic Conditioning in LLM Next-Location Prediction

Large language models (LLMs) are increasingly used for individual next-location prediction, while sociodemographic conditioning is common in LLM-based travel simulation. Yet the incremental predictive value of sociodemographic attributes remains unclear. To directly test this contribution, sociodemographic records were linked with passively sensed mobility data from 5,000 Shenzhen residents to construct a closed-set benchmark in which models rank 100 candidate destinations. Each prediction instance is evaluated with and without age, gender, occupation and income, while holding mobility history, candidates and all other prompt content fixed. Results show that across four history lengths, the paired change in top-1 accuracy ranges from -0.8 to +0.5 percentage points, with no detectable gain from attributes. This result remains consistent when stay history is withheld, across alternative prediction times, in two additional LLMs and in a supervised reranker trained on the same benchmark. The null does not reflect a lack of model responsiveness to demographic information, as permuted attributes reduce LLM accuracy whereas correctly matched attributes do not improve it. A further asymmetry emerges in the reverse predictive direction, as pre-cut mobility trajectories recover income with an AUC of 0.708, while sociodemographic attributes contribute little to next-location prediction. Beyond demographic conditioning, candidate construction exerts a much larger influence on reported performance. Removing distance raises top-1 accuracy by 7.7 percentage points under proximity sampling but lowers it by 22.3 points under popularity sampling, with the reversal reproduced across all three LLMs. These results distinguish demographic association from incremental predictive usefulness and show that sampled next-location accuracy depends strongly on how candidate alternatives are constructed.

cs.CY↗

arXiv2026

The Evolution of Thermal and Non-thermal Emission Components in GRB 250920C

We present a temporal and time-resolved spectral analysis of GRB 250920C using \textit{Fermi}/GBM and \textit{Swift}/BAT data. The prompt emission consists of two distinct episodes, EI and EII, separated by a significant quiescent interval. We perform Bayesian spectral fitting with empirical, thermal, composite, and physical synchrotron models. In EI, the spectra show clear evidence for an additional thermal component. The BB+PL model is preferred in most bright time bins, and joint \textit{Fermi}/GBM+\textit{Swift}/BAT fits further support the presence of this component. The blackbody temperature generally decreases with time, the blackbody flux follows the pulse profile, and the thermal flux fraction remains high. Fireball-parameter estimates give a photospheric Lorentz factor of a few hundred and an initial radius of $r_{0}\sim10^{9}$--$10^{10}$ cm, with $1+σ_{0}$ of order unity and $η\gg1$, supporting a thermally dominated baryonic outflow. In contrast, EII is dominated by non-thermal emission. Its low-energy photon indices do not significantly exceed the synchrotron line of death, and the spectra are well described by fast-cooling synchrotron radiation in a decaying magnetic field. The magnetization constraint gives $σ_{\min}\sim1.1$--$4.3$. Since these values exceed unity, they suggest that EII may be Poynting-flux dominated. These results therefore suggest a possible transition in GRB 250920C from a fireball-dominated EI phase to a Poynting-flux-dominated EII phase.

astro-ph.HE↗

arXiv2026

AIBuildAI-2.5: Efficient Autonomous AI Model Development Through LLM-Guided Tree Search

Autonomous agents that automatically build artificial intelligence (AI) models could broaden access to AI across science and engineering. A popular line of such agents frames model building as a code search problem and solves it by tree search, in which each node is a candidate program and the tree grows by generating a child program from a parent, and these agents now approach the capability of experienced AI engineers on realistic benchmarks. However, these agents have three weaknesses in efficiency that have not been fully addressed. First, only a small number of candidates can be executed within a realistic budget, so search rules that rank nodes by executed rewards, such as Monte Carlo-style tree search, rely on few and noisy scores and select the next node to explore less effectively. Second, no resource-aware strategy is used to schedule training jobs, which can lower hardware utilization and training efficiency. Third, every agent call is served by a single powerful model, which inflates inference cost. Here we introduce AIBuildAI-2.5, an agentic system that carries out the tree search with LLM agents and addresses each of the three issues. AIBuildAI-2.5 proposes a novel LLM-guided tree search, in which a judge scores each candidate on its expected improvement, grounding, and feasibility, and a selector ranks the pool of candidates from these scores and the state of the search. In addition, AIBuildAI-2.5 comprises a scheduler that launches training jobs with the current hardware resource status taken into account and a router that assigns lower-cost LLMs to less demanding tasks while reserving the most capable LLM for the most challenging sub-tasks in the AI model building workflow. AIBuildAI-2.5 ranks first on MLE-Bench with a medal rate of 73.3%, and outperforms a strong baseline on six autonomous AI research tasks from AIRS-Bench.

cs.CL↗

arXiv2026

HAT-4D: Lifting Monocular Video for 4D Multi-Object Interactions via Human-Agent Collaboration

Extracting dynamic 4D object interactions from massive, in-the-wild monocular videos offers a highly efficient data collection pathway for scaling Embodied AI and training VLAs. However, existing monocular 4D reconstruction methods primarily focus on isolated objects, often failing under the severe occlusions and complex dynamics inherent in multi-object interactions. To bridge this gap, we propose HAT-4D, the first agentic framework designed to reconstruct the 3D geometry, temporal dynamics, and physical interactions of multiple objects from a single video. By integrating VLMs with a multi-level human-in-the-loop feedback mechanism, HAT-4D efficiently resolves depth ambiguities and interaction-induced occlusions during 3D generation and 4D propagation, yielding physically plausible assets without relying on expensive multicamera rigs. As a scalable data engine, HAT-4D facilitates the creation of MVOIK-4D, an open-world benchmark for monocular 4D interaction reconstruction, accompanied by a novel multi-dimensional evaluation protocol focused on physical plausibility and temporal consistency. Extensive experiments demonstrate that HAT-4D achieves SOTA performance on most evaluation metrics, while maintaining competitive semantic alignment. Ablation studies show that introducing a small amount of human feedback improves interaction reconstruction. Moreover, the data produced by HAT-4D effectively improves baseline performance when used for fine-tuning. Our data and code are available at https://lijiaxin0111.github.io/HAT4D/

cs.CV↗

arXiv2026

PASTEL: Panoramic Alignment for Monocular 4D Scene Reconstruction

Reconstructing 4D scenes from casually captured monocular video is vital for applications in virtual reality (VR) and embodied AI. Recent advances in 4D reconstruction and novel view synthesis have substantially propelled this capability. However, existing reconstruction methods generally cannot recover regions beyond visible camera limits. Consequently, we introduce a new paradigm that achieves 4D scene synthesis by combining visible-region reconstruction from monocular input with invisible-region generation beyond observable camera boundaries. We present Panoramic Alignment for Strategic Exploitation of Generative Priors (PASTEL). Specifically, PASTEL proposes panoramic scene alignment, a novel representation that reformulates the intractable 3D "invisible region" exploration into a tractable 2D directional trajectory planning. This is achieved by reducing the viewpoint planning from 6-DoF search to a 2D directional search with explicit visibility boundaries. By operating within this panoramic space, our method strategically identifies camera trajectories that maximize exploration beyond observable boundaries while minimizing viewpoint deviation. Experimental results show that PASTEL can not only extrapolate plausible scene content beyond the observable boundaries of input monocular videos, but also substantially boost monocular 4D reconstruction performance. PASTEL outperforms the previous state-of-the-art method by 0.9dB in full-image PSNR on the DyCheck IPhone dataset.

cs.CV↗

arXiv2026

Gradient-based Model Shortcut Detection for Time Series Classification

Deep learning models have attracted lots of research attention in time series classification (TSC) task in the past two decades. Recently, deep neural networks (DNN) have surpassed classical distance-based methods and achieved state-of-the-art performance. Despite their promising performance, deep neural networks (DNNs) have been shown to rely on spurious correlations present in the training data, which can hinder generalization. For instance, a model might incorrectly associate the presence of grass with the label ``cat" if the training set have majority of cats lying in grassy backgrounds. However, the shortcut behavior of DNNs in time series remain under-explored. Most existing shortcut work are relying on external attributes such as gender, patients group, instead of focus on the internal bias behavior in time series models. In this paper, we take the first step to investigate and establish point-based shortcut learning behavior in deep learning time series classification. We further propose a simple detection method based on other class to detect shortcut occurs without relying on test data or clean training classes. We test our proposed method in UCR time series datasets.

cs.LG↗

arXiv2026

TacPAC: Tactile Prediction and Real-Time Action Correction in World-Action Models for Contact-Rich Manipulation

World-action models guide action generation with predicted future observations, but vision-centric predictions miss the local contact cues that decide contact-rich manipulation. However, naively predicting future tactile observations as additional views recovers only a third of the achievable gain in our experiments. This gap reflects a timing mismatch: predictions precede execution, while tactile feedback arrives during it. We introduce TacPAC, which turns tactile prediction into real-time action correction. Once the base model has planned an action chunk, TacPAC caches the predicted contact that plan was conditioned on together with the plan's own representation, and a tactile expert reads each newly observed tactile image against that cache to correct the actions not yet executed. Feedback is thus interpreted against what the plan anticipated rather than in isolation, and one correction is a single pass over that cache, $20.7\times$ cheaper than regenerating the chunk. On five real-robot tasks spanning precision insertion, fragile-object handling, object reorientation, and long-horizon manipulation, TacPAC leads every task and raises the average from 22% for its vision-only base model to 64%. Code is available at https://github.com/LogosRoboticsGroup/TacPAC.

cs.RO↗

arXiv2026

Towards Robust Driving Perception: A Flexible Scale-Driven Family for Self-Supervised Monocular Depth Estimation

Self-Supervised Monocular Depth Estimation (MDE) has garnered attention in recent years due to its independence from ground truth. However, most existing models are limited to a single scale and exhibit considerable performance degradation in complex driving environments. Networks specifically designed to handle dynamic traffic participants tend to be overly complex, hindering their deployment on resource-constrained automotive edge devices. To address these limitations and move towards robust driving perception, we propose FlexDepth, a scale-driven and flexible family of self-supervised MDE models tailored for challenging road scenarios. FlexDepth employs a two-stage static-dynamic decoupled training strategy, enabling the independent assessment of confidence for both static backgrounds and dynamic road objects. Furthermore, it introduces a meticulously designed Scale-Driven Decoder (SDD) to dynamically select components based on scale size, facilitating efficient feature fusion and the output of high-precision depth maps. Extensive experiments on standard driving benchmarks demonstrate that without any auxiliary information, our model achieves state-of-the-art performance across arbitrary scales with minimal computational overhead. Our smallest model, Flex-Nano, requires only 0.7 GFLOPs and achieves 37.6 FPS on mobile platforms, ensuring reliable real-time perception while maintaining excellent zero-shot generalization. Our source code is available: https://github.com/startnew/flexdepth

cs.CV↗