Searcharxiv⌕ Search

arXiv subjects

Hao Zhang

Publications and source records attributed to Hao Zhang.

At least 109 records · Page 6Linked to original sources
arXiv2026

SiZeUp: Fast 3D Proxy from Aerial Images via Depth Ordinal Loss

We present SiZeUp, a fast and scalable approach for constructing large-scale 3D urban proxy models directly from calibrated oblique aerial imagery. Our method adopts a height-from-footprint representation, reducing 3D building abstraction to a low-dimensional optimization problem in which building footprints are extruded by a single height parameter. To enable efficient and robust height estimation, we introduce an ordinal depth consistency loss that enforces agreement between the relative depth ordering of rendered proxies and depth priors predicted by a monocular depth model. This is realized through a differentiable renderer that maps parametric building proxies into multi-view depth images, allowing gradients to be propagated from depth supervision to building heights. Our ordinal formulation produces stable optimization in practice and avoids explicit feature matching or dense point cloud reconstruction. Rather than relying on metric depth, which can be unreliable under monocular scale ambiguity, our ordinal depth consistency loss operates on relative depths, providing a more reliable signal across views. Combined with an efficient dynamic view selection, our approach achieves a 23-52$\times$ speedup over state-of-the-art proxy reconstruction pipelines while maintaining comparable proxy-level coverage and volume consistency, making it well suited for large-scale urban modeling tasks.

cs.CV↗
arXiv2026

Fine-tuning an ECG Foundation Model to Predict Coronary CT Angiography Outcomes

Coronary artery disease (CAD) remains a major global public health burden, yet scalable pre-imaging risk stratification tools are limited. In this multicenter study, we developed and validated an artificial intelligence-enabled electrocardiography (AI-ECG) model using coronary computed tomographic angiography (CCTA) as the anatomical reference to predict vessel-specific hemodynamically significant stenosis ($\geq 70\%$ for RCA, LAD, LCX; $\geq 50\%$ for LM). The model was evaluated in internal and external cohorts, clinically normal ECGs, and prespecified demographic and clinical subgroups. It showed discrimination across vessels in internal validation and consistent external and normal ECG performance. Predicted probabilities increased with CCTA-defined stenosis severity and were converted into vessel-specific low-, intermediate-, and high-risk strata. Calibration and decision curve analyses supported its clinical utility. Integration with guideline-based pre-test probability improved risk reclassification, enhanced rule-out performance, and reduced the gray-zone proportion. In longitudinal follow-up, model-defined risk groups showed clear separation in major adverse cardiovascular events. Waveform- and attribution-based analyses identified structured ECG differences and physiologically meaningful signal regions linked to high-risk predictions. These results support AI-ECG as a feasible tool for pre-imaging risk stratification and clinical triage, warranting prospective validation in broader clinical settings.

cs.CV↗
arXiv2026

ProtoFlow: Mitigating Forgetting in Class-Incremental Remote Sensing Segmentation via Low-Curvature Prototype Flow

Remote sensing segmentation in real deployment is inherently continual: new semantic categories emerge, and acquisition conditions shift across seasons, cities, and sensors. Despite recent progress, many incremental approaches still treat training steps as isolated updates, which leaves representation drift and forgetting insufficiently controlled. We present ProtoFlow, a time-aware prototype dynamics framework that models class prototypes as trajectories and learns their evolution with an explicit temporal vector field. By jointly enforcing low-curvature motion and inter-class separation, ProtoFlow stabilizes prototype geometry throughout incremental learning. Experiments on standard class- and domain-incremental remote sensing benchmarks show consistent gains over strong baselines, including up to 1.5-2.0 points improvement in mIoUall, together with reduced forgetting. These results suggest that explicitly modeling temporal prototype evolution is a practical and interpretable strategy for robust continual remote sensing segmentation. Open-source code:https://github.com/dudududke/protoflow.

cs.CV↗
arXiv2026

A Model-Internal Protocol for Assessing Multimodal Models as Integrated Systems

As Large Vision-Language Models increasingly aim to integrate visual generation and understanding within a single parameter space, evaluating such structural unification in a cohesive manner remains a critical challenge. Current evaluation protocols predominantly treat generative and discriminative capabilities as separate tasks, leaving a gap in system-level evaluation for unified multimodal models (UMMs). In this work, we propose Self-Generative-Understanding (SGU), a novel, annotation-free evaluation framework that probes the integrated capabilities of unified models through a semantic closed-loop challenge. Without requiring new annotations, SGU leverages the dual understanding-and-generation abilities of UMMs by asking them to first perceive an image and produce a textual description, subsequently reconstruct a visual context based on that description, and finally perform reasoning over the self-generated output. This pipeline provides a zero-cost testbed that yields an integrated performance score specifically tailored for evaluating UMMs as unified systems. Extensive experiments show that even high-performing UMMs often struggle to reason over their own generated contexts, revealing limitations that are not captured by separate evaluations of understanding or generation alone. Our work provides a complementary holistic evaluation framework and offers a foundation for benchmarking the development of next-generation unified multimodal models.

cs.CV↗
arXiv2026

Large Scale Entanglement Structure Detection in 100-Qubit Systems via Local Joint Measurements

Identifying the entanglement structure of a many-body quantum state, namely how its constituents partition into unentangled blocks, is a central task in quantum information science, yet conventional tomography scales exponentially with system size. Here we introduce a scalable framework that recognizes large-scale entanglement structures directly from local correlation fingerprints. By choosing a representative local Pauli basis that satisfies a boundary-matching condition p_1 = p_R, the entire chain is read out in a single measurement configuration, keeping the measurement effort independent of system size. In noisy simulations, this single-basis protocol classifies GHZ-, W-, and cluster-type structures among 30 candidate partitions with a mean accuracy exceeding 95% for systems of up to 100 qubits. We further validate the protocol on a superconducting quantum processor, where it reliably classifies block structures for systems of up to 13 qubits before noise- and depth-induced degradation sets in at larger sizes. By mapping these failure modes explicitly, our results delineate the boundary of hardware-level scalability and point to a concrete strategy for characterizing entanglement structure on near-term quantum devices.

quant-ph↗
arXiv2026

Ab initio-based Deep-Learning Prediction of Carrier Mobility in Strongly Anharmonic Materials

Predicting charge transport in strongly anharmonic materials, particularly ultralow thermal conductors, remains a major challenge for first-principles methods. In such systems, perturbative treatments of electron-phonon interactions and the harmonic phonon picture often break down, necessitating non-perturbative approaches. The ab initio Kubo-Greenwood(aiKG) formalism provides a rigorous framework for evaluating temperature-dependent carrier transport beyond the harmonic approximation. Nevertheless, its practical application is computationally demanding because it requires large supercells, extensive statistical sampling, and extrapolation to the zero-frequency limit. In this work, we introduce an artificial-intelligence(AI)-assisted aiKG framework that incorporates the deep-learning Hamiltonian model. By predicting the Kohn-Sham Hamiltonian with sub-meV accuracy for supercells of up to 250 atoms, the model bypasses the costly iterative self-consistent field calculations while retaining first-principles reliability within the scope of effects captured by the training data. Using a strongly anharmonic thermal insulator, potassium iodide(KI) as a benchmark system, we demonstrate that the proposed approach enables efficient simulations of electronic structure and transport properties from a large supercell. The framework reproduces temperature-dependent carrier mobilities, spectral functions, and effective masses in close agreement with the underlying density functional theory while reducing computational cost to 10%. These results suggest that the AI-assisted aiKG framework can make non-perturbative transport calculations tractable for strongly anharmonic materials, opening a scalable route towards realistic simulations and accelerated discovery of new functional materials.

cond-mat.mtrl-sci↗
arXiv2026

Evaluating the Diversity of AI-Generated Content with Diversity Profiles

Diversity is a fundamental criterion for evaluating generative artificial intelligence (AI) systems, yet its measurement remains inherently ambiguous. Existing approaches typically represent generated samples in an embedding space, compute pairwise distances or similarities, and aggregate them into a single scalar score. Such scalar summaries are convenient, but they often encode different inductive biases and may yield contradictory rankings of the same sample sets. In this paper, we argue that diversity evaluation for AI-generated content is intrinsically under-specified when reduced to a single number. We first review representative diversity metrics, and then diagnose their limitations from two complementary perspectives: an axiomatic analysis showing that no representative scalar metric satisfies all desirable properties simultaneously, and an empirical analysis showing that high-dimensional representation spaces can induce concentrated, modality-dependent distance distributions. To address these issues, we propose diversity profiles: curve-valued, condition-aware summaries that evaluate a parameterized diversity family across a range of thresholds, scales, exponents, or orders under a specified representation and distance or kernel function. Diversity profiles reveal whether a comparison is robust across resolutions or instead depends on an arbitrary parameter choice. We instantiate profiles for several representative metric families and demonstrate their practical use in generative AI evaluation. Overall, diversity profiles provide a more transparent and resolution-aware framework for comparing the diversity of AI-generated content.

cs.AI↗
arXiv2026

Learning Versatile Humanoid Manipulation with Touch Dreaming

Humanoid robots promise general-purpose assistance, yet real-world humanoid loco-manipulation remains challenging because it requires whole-body stability, end-effector dexterity, and contact-aware interaction under frequent contact changes. In this work, we study dexterous, contact-rich humanoid loco-manipulation. We first develop an RL-based lower-body controller that serves as the stability backbone for whole-body execution during complex manipulation. Building on this controller, we develop a VR-based whole-body humanoid data collection system that integrates dexterous hands and tactile sensing for contact-rich manipulation. We then propose Humanoid Transformer with Touch Dreaming (HTD), a multimodal encoder-decoder Transformer that models touch as a core modality alongside multi-view vision and proprioception. HTD is trained in a single stage with behavioral cloning augmented by touch dreaming: in addition to predicting action chunks, the policy predicts future hand-joint forces and future tactile latents, with tactile-latent targets provided by an exponential moving average target encoder without requiring a separate tactile pretraining stage. This encourages the policy to learn contact-aware representations for dexterous manipulation. Across five real-world contact-rich tasks, HTD achieves a 90.9% relative improvement in average success rate over the stronger baseline for each task. Ablation results further show that latent-space tactile prediction is more effective than raw tactile prediction, yielding a 30% relative gain in success rate. These results demonstrate that our touch-dreaming-enhanced learning system enables versatile, high-dexterity humanoid manipulation in the real world. More information and open-source materials are available at humanoid-touch-dream.github.io.

cs.RO↗
arXiv2026

D2-ScaleAgent: Dual-Dimensional Scaling for Long Document Understanding

Multi-modal retrieval-augmented generation (RAG) is a key technique for visually rich long document understanding. Existing multi-modal RAG methods are progressively advancing toward multi-agent systems: they first retrieve relevant pages based on a query, and then iteratively understand information within those pages. However, these methods typically rely on fixed workflows and lack the ability to dynamically scale computation at test time, often leading to insufficient evidence. To address this, we propose D2-ScaleAgent, an agentic framework that introduces a dual-dimensional scaling paradigm for retrieval and reasoning. The core of D2-ScaleAgent is a Verifier agent-driven dynamic routing loop based on the intrinsic difficulty of the query, centered around a continuously updated evidence bank that serves as the agent's dynamic working memory: when retrieval needs to be expanded, the agent routes outward (retrieval scaling), decomposing the query into attributes and performing parallel page retrieval, followed by adaptive pruning to ensure comprehensive evidence coverage. When fine-grained reasoning is required, the agent routes inward (reasoning scaling), dynamically selecting sub-agents with varying granularity and count to extract evidence from pages. Finally, D2-ScaleAgent achieves logical closure over the evidence chain. Extensive experiments demonstrate that D2-ScaleAgent is effective on long and visually rich document benchmarks like MMLongBench-Doc, LongDocURL, etc.

cs.CL↗
arXiv2026

Nearly permanental cospectral graphs

Let $G$ be a simple graph of order $n$ with adjacency matrix $A= (a_{ij})$. The \emph{determinant} and the \emph{permanen}t of the matrix $A$ are defined as \[\mathrm{det}A= \sum_{σ\in S_n}\mathrm{sgn}(σ) \prod_{i=1}^n a_{iσ(i)}\quad\text{and}\quad\mathrm{per}A= \sum_{σ\in S_n} \prod_{i=1}^n a_{iσ(i)},\]respectively. The polynomials $ϕ(G;x) =\mathrm{det}(xI-A(G))$ and $π(G;x) =\mathrm{per}(xI-A(G))$ are called the \emph{characteristic polynomial} and the \emph{permanental polynomial} of $G$, respectively. Two graphs are said to be \emph{nearly cospectral} with respect to the determinant (resp. permanent) if the difference of their characteristic (resp. permanental) polynomials is a constant. Lv et al. introduced the nearly cospectral graphs problem with respect to the determinant, and provided partial results in the case modulo 4. In this paper, we mainly prove that the corresponding results also hold for the nearly cospectral graphs problem with respect to the permanent. The determinant and permanent are the immanants corresponding to the irreducible characters $(1^n)$ and $(n)$ of the symmetric group $ S_n $, respectively. Here, the \emph{immanant} $d_λ(A)$ of $A$ is defined as \[d_λ(A) = \sum_{σ\in S_n} χ_λ(σ) \prod_{i=1}^n a_{iσ(i)},\] where $χ_λ$ is the irreducible character of $ S_n $ indexed by the partition $ λ$. The immanantal polynomial of $G$ associated with $ χ_λ$ is given by $ ϕ_λ(G;x)=d_λ(xI-A) $. In this paper, we also establish a similar result for nearly immanantal cospectral graphs in $\mathbb{F}_2[x]$ for all irreducible characters $χ_λ$.

math.CO↗
arXiv2026

Large values of quadratic character sums

In this paper, we investigate large values of quadratic Dirichlet character sums. We prove new Omega results for both short and long quadratic character sums under the assumption of the Generalized Riemann Hypothesis (GRH), which improve the previous results.

math.NT↗
arXiv2026

Temporal Logic Guided Universal Task Representations for Reinforcement Learning

Task guided agents demonstrate strong performance in a wide range of complex tasks. However, most existing task representation algorithms are tailored to specific contexts and struggle to generalize across diverse scenarios. Moreover, they typically depend on gradient signals from reinforcement learning controllers to update their weights, which can degrade both representation quality and learning efficiency. To overcome these limitations, we propose LOTUS, a temporal logic inspired universal task representation framework that can be seamlessly integrated into any RL algorithm to enhance agent performance across diverse task settings. Specifically, we design a novel task representation architecture capable of modeling relationships and extracting task semantics from LTL formulas. We further introduce a more effective update mechanism that treats the LTL encoder as a policy, thereby improving representation capacity. To enhance stability and robustness, LOTUS leverages the bisimulation metric, which provides theoretical guarantees for LTL representation, including behavioral equivalence, optimality fidelity, and trajectory robustness. Experimental results show that LOTUS outperforms most existing methods in learning efficiency, generalization capability, and representation quality. Specifically, LOTUS accelerates convergence over 20% in single-task scenarios, achieves a 15%-45% higher success rate in unseen manipulation tasks, and improves generalization performance over 25% in complex multi-task environments with increased sub-goal depth or conjunctions. The corresponding code, videos, and appendix are available at: https://lotus-website.github.io/.

cs.RO↗
arXiv2026

Stochastic Liouville-transport theory of light-atom interaction noise in thermal atomic vapors

Atom-light interaction noise can limit thermal-vapor sensing. Existing theories often treat internal-state dynamics, finite-mode atomic motion, and stochastic renewal separately, obscuring their coupled contributions to measured noise. We develop a general stochastic Liouville-transport theory, tested against polarization-resolved resonant Cs D$_2$ spectra. Joint experiment-theory analysis identifies atom-light noise below approximately 100 kHz as transit-dominated. Ballistic motion through the finite Gaussian mode modulates both the coupling-weighted effective atom number and trajectory-dependent Rabi coupling, producing predominantly common-mode noise. Boundary renewal introduces atoms with independently sampled ground-state sublevels, generating differential population fluctuations with opposite effects on the circular channels. Under an applied longitudinal magnetic field, experiment and theory show the same qualitative nonmonotonic change in common-mode suppression, supporting Zeeman redistribution of the channel responses. The framework can analyze noise in other thermal-atom sensors, including Rydberg-atom electric-field measurements.

quant-ph↗
arXiv2026

PhyAI: Real-Time Physical AI at the Edge, Scalable Rollouts in the Cloud

Physical AI policies require inference throughout their lifecycle, including model evaluation, cloud reinforcement learning rollout, edge GPU serving, and onboard deployment. Although these settings share the same checkpoint and action semantics, they often rely on separate inference programs. To unify them, we build PhyAI, a Physical AI inference engine with a single runtime that keeps architecture-specific conditioning, solver, cache, and output logic in model adapters while sharing graph execution, kernels, memory management, and parallel services. The same codebase runs vision-language-action (VLA) models and world-action models (WAMs) on single or multiple GPUs across onboard, edge, and cloud deployments. We used the adapter interface to add MiniCPM-Robot on the day of its release. PhyAI achieves 1.40x-4.65x speedups over the official implementations of pi0, pi0.5, GR00T N1.7, and MiniCPM-Robot. On Cosmos3-Nano-Policy-DROID it reduces latency from 2.46 to 1.18 s on eight H20 GPUs (CFG=2, TP=4), a 2.08x speedup. Specialized runtimes remain faster in several configurations, so our goal is one runtime with competitive latency rather than the fastest result in every case. Detailed profiles reveal why different models need different execution policies: on a Hopper-series GPU at batch size one, the pi0.5 action expert accounts for 8.8% of FLOPs but 57.2% of latency; at batch size 32 its share drops to 13.5% and throughput reaches about 100 samples/s. Cosmos3 remains generation-dominated and gains only 14.3% throughput as batch size increases from 1 to 16. We further introduce the control-time Roofline, which distinguishes inference-bound from environment-bound control; the measured pi0.5 points on four LIBERO suites are environment-bound while Cosmos3 stays inference-bound. Code and benchmarks: https://github.com/mingti-org/phyai.

cs.AI↗
arXiv2026

Illusion or Integrity? Geometrical Consistency Metric for AIGC Video Quality Evaluation

Recently, AI-driven video generation has attracted considerable attention. This surge increases the demand for reliable video quality assessment (VQA) metrics to evaluate AI-generated content (AIGC) videos and guide model optimization. Existing studies assess video quality through visual harmony, video-text consistency, and domain-specific alignment, yet lack quantitative metrics for measuring fidelity to physical laws. To address this limitation, we present a novel benchmark that evaluates the quality of AIGC videos based on their compliance with physical principles by quantitatively measuring geometric consistency across frames extracted from generated sequences. This serves as a proxy for estimating the extent to which generated videos conform to real-world physical rules. Specifically, GeoCon-Bench captures global motion through translation estimation, fits homography or fundamental matrix models using background correspondences, and reports complementary metrics, including inlier ratio and geometric error. We also release a dataset containing 20 scenes across six motion categories. Experiments on state-of-the-art AIGC models demonstrate the reliability of GeoCon-Bench as a video quality assessment metric.

cs.CV↗
arXiv2026

Discovery of four Narrow-line Type-1 Quasars at z = 2.54 -- 6.06 with Subaru 'Onohi'ula PFS-SSP

We report the discovery of four narrow-line type-1 quasars (NLQ1s) from the Subaru 'Onohi'ula Prime Focus Spectrograph Subaru Strategic Program (PFS-SSP). To extend a comprehensive understanding of this population to the distant universe, we searched for broad-line AGNs (quasars) whose rest-UV spectral shapes resemble those of local NLS1 galaxies, utilizing the Ly$α$ line profile. We identified four NLQ1s, including one candidate, at $z=2.54$ -- $6.06$. The obtained spectra show a prominent Ly$α$ emission line with a very low continuum level. The measured full width at half maximum (FWHM) of Ly$α$ ranges from $1410 \ \mathrm{km\,s^{-1}}$ to $1510 \ \mathrm{km\,s^{-1}}$ with a mean of $\sim 1430 \,\mathrm{km\,s^{-1}}$. Although we used only line width information for the NLQ1 selection, the resulting equivalent widths (EW) of Ly$α$ are higher (the mean $\sim 190$\,Å\ in the rest frame) than the typical values of quasars. Two quasars have exceptionally large Ly$α$ luminosity, accounting for $\sim 3$ -- $4$\% of the bolometric luminosity. We measured black hole masses and Eddington ratios for three objects whose C\,\textsc{iv} line is available. The resulting black hole masses are $\sim 10^{6.9}$ -- $10^{7.7}\, M_{\odot}$ with high Eddington ratios ($> 0.1$), consistent with those of NLS1 galaxies reported in the low-$z$ universe. N\,\textsc{v} and C\,\textsc{iv} broad absorption lines are detected in one object. This quasar can be classified as a mini-BAL quasar, which further supports the idea that this NLQ1 is indeed a luminous version of an NLS1 galaxy.

astro-ph.GA↗
arXiv2026

A JEPA-Based Field-Layer World Model for Bridging Channel Prediction and Estimation

Channel state information (CSI) acquisition, reconstruction, and prediction are fundamental yet costly tasks in modern MIMO-OFDM wireless systems. Direct coefficient-level prediction of raw CSI is fragile in realistic propagation environments, since small spatial perturbations, local scattering changes, and phase variations can cause large errors in the complex channel domain. However, the underlying wireless propagation field still contains stable and predictable structures that can be exploited across time, frequency, antenna, and carrier dimensions. Motivated by this observation, we propose a JEPA-based field-layer world model (FWM) that learns a shared latent propagation state from multi-resolution CSI observations across the considered carrier bands and predicts its task-relevant evolution in the latent domain. The proposed FWM maps multiple CSI observation resolutions to a shared latent propagation-field space through scale-specific tokenizer heads. A latent prediction backbone is then trained to infer masked or future field states, while an incremental multi-scale alignment strategy allows new observation scales to be incorporated without retraining the entire model from scratch. For downstream reconstruction, the predicted latent field is used as a structured prior and combined with sparse current pilots. Experiments on single-band and cross-band reconstruction demonstrate improved symbol detection and, more notably, substantial beamforming gains despite modest NMSE improvements, indicating that FWM captures task-relevant spatial propagation structure beyond coefficient-wise CSI fitting.

eess.SP↗
arXiv2026

Attn-QAT: 4-Bit Attention With Quantization-Aware Training

Achieving reliable 4-bit attention is a prerequisite for end-to-end FP4 computation on emerging FP4-capable GPUs, yet attention remains the main obstacle due to FP4's tiny dynamic range and attention's heavy-tailed activations. This paper presents the first systematic study of 4-bit quantization-aware training (QAT) for attention. We find ``drop-in'' QAT -- naively combining an FP4 forward pass with high-precision Flash Attention (FA)-style backward pass -- leads to training instability. We identify two key principles for stable FP4 attention: (1) matching low-precision recomputation of attention scores in the backward pass and (2) resolving implicit precision assumptions in FA's gradient calculation. Based on these insights, we propose Attn-QAT and implement fused Triton kernels for training and FP4 inference kernels. Across diffusion and language models, Attn-QAT recovers the quality drop from FP4 attention without explicit outlier-mitigation heuristics used in prior FP4 attention, and delivers up to a 1.5x speedup on an RTX 5090 over SageAttention3 and up to a 1.74x speedup over FA4 on a GB300. Video demos can be found at https://drive.google.com/drive/folders/190F6xbBDUF2kGQYIcXBt3ehSYij5jlim?usp=sharing. Code can be found at https://haoailab.com/FastVideo/training/attn_qat/.

cs.LG↗