Searcharxiv⌕ Search

arXiv subjects

Yu Zhang

Publications and source records attributed to Yu Zhang.

At least 307 records · Page 17Linked to original sources
arXiv2026

Nonclassicality of fully degenerate triple-photon states and its application in generating non-Gaussian entanglement

We theoretically demonstrate via numerical modeling that fully degenerate triple-photon states generated by three-mode spontaneous parametric down-conversion can be categorized into four distinct states: 0-phase, $π$/2-phase, $π$-phase, and 3$π$/2-phase squeezed states. Using quantum relative entropy and Wigner negativity as quantitative measures, we show that the nonGaussianity and nonclassicality of these squeezed states increase with the increase of interaction strength. Analogous to Gaussian scenarios, these squeezed states serve as basic building blocks for deterministic preparation of two-mode non-Gaussian states. We study the correlation properties of two-mode state generated by interfering 0-phase and $π$-phase squeezed states on a beam splitter, and reveal its entanglement at 3rd and 6th-order moments using the positive partial transposition criterion based on higher-order covariance matrices. In particular, by adjusting the intensities of the two input beams, the entanglement of the two output modes at the 3rd- and 6th-order moments can be dynamically modulated. Our results highlight the nonclassical nature of fully degenerate triple-photon states and establish a pathway for preparing non-Gaussian entanglement based on such states.

quant-ph↗
arXiv2026

AgenticRS-EnsNAS: Ensemble-Decoupled Self-Evolving Architecture Search

Neural Architecture Search (NAS) deployment in industrial production systems faces a fundamental validation bottleneck: verifying a single candidate architecture pi requires evaluating the deployed ensemble of M models, incurring prohibitive O(M) computational cost per candidate. This cost barrier severely limits architecture iteration frequency in real-world applications where ensembles (M=50-200) are standard for robustness. This work introduces Ensemble-Decoupled Architecture Search, a framework that leverages ensemble theory to predict system-level performance from single-learner evaluation. We establish the Ensemble-Decoupled Theory with a sufficient condition for monotonic ensemble improvement under homogeneity assumptions: a candidate architecture pi yields lower ensemble error than the current baseline if rho(pi) < rho(pi_old) - (M / (M - 1)) * (Delta E(pi) / sigma^2(pi)), where Delta E, rho, and sigma^2 are estimable from lightweight dual-learner training. This decouples architecture search from full ensemble training, reducing per-candidate search cost from O(M) to O(1) while maintaining O(M) deployment cost only for validated winners. We unify solution strategies across pipeline continuity: (1) closed-form optimization for tractable continuous pi (exemplified by feature bagging in CTR prediction), (2) constrained differentiable optimization for intractable continuous pi, and (3) LLM-driven search with iterative monotonic acceptance for discrete pi. The framework reveals two orthogonal improvement mechanisms -- base diversity gain and accuracy gain -- providing actionable design principles for industrial-scale NAS. All theoretical derivations are rigorous with detailed proofs deferred to the appendix. Comprehensive empirical validation will be included in the journal extension of this work.

cs.LG↗
arXiv2026

HalluClean: A Unified Framework to Combat Hallucinations in LLMs

Large language models (LLMs) have achieved impressive performance across a wide range of natural language processing tasks, yet they often produce hallucinated content that undermines factual reliability. To address this challenge, we introduce HalluClean, a lightweight and task-agnostic framework for detecting and correcting hallucinations in LLM-generated text. HalluClean adopts a reasoning-enhanced paradigm, explicitly decomposing the process into planning, execution, and revision stages to identify and refine unsupported claims. It employs minimal task-routing prompts to enable zero-shot generalization across diverse domains, without relying on external knowledge sources or supervised detectors. We conduct extensive evaluations on five representative tasks-question answering, dialogue, summarization, math word problems, and contradiction detection. Experimental results show that HalluClean significantly improves factual consistency and outperforms competitive baselines, demonstrating its potential to enhance the trustworthiness of LLM outputs in real-world applications.

cs.CL↗
arXiv2026

Machine Learning Reconstruction of High-Dimensional Electronic Structure from Angle-Resolved Photoemission Spectroscopy

The emergent behavior of quantum materials is governed by their electronic structure, which can be experimentally probed by photoemission spectroscopy techniques that generate a four-dimensional dataset of energy and momentum. However, the quantitative extraction of Hamiltonian parameters from these high-dimensional spectra remains a significant challenge, currently relying on labor-intensive, expert-dependent analysis rather than standardized workflows. Here, we introduce a deep learning framework based on implicit neural representations to accelerate the retrieval of Hamiltonian parameters in two types of transition-metal oxides: perovskite nickelates and manganites. Our approach outperforms traditional analytical fitting procedures, yielding superior agreement with experimental Fermi surface topologies and energy-momentum dispersions. This work highlights the potential of deep learning tools to bridge the gap between theory and experiment, paving the way for high-throughput, autonomous discovery pipelines in quantum materials.

cond-mat.str-el↗
arXiv2026

Topology-Guided Biomechanical Profiling: A White-Box Framework for Opportunistic Screening of Spinal Instability on Routine CT

Routine oncologic computed tomography (CT) presents an ideal opportunity for screening spinal instability, yet prophylactic stabilization windows are frequently missed due to the complex geometric reasoning required by the Spinal Instability Neoplastic Score (SINS). Automating SINS is fundamentally hindered by metastatic osteolysis, which induces topological ambiguity that confounds standard segmentation and black-box AI. We propose Topology-Guided Biomechanical Profiling (TGBP), an auditable white-box framework decoupling anatomical perception from structural reasoning. TGBP anchors SINS assessment on two deterministic geometric innovations: (i) canal-referenced partitioning to resolve posterolateral boundary ambiguity, and (ii) context-aware morphometric normalization via covariance-based oriented bounding boxes (OBB) to quantify vertebral collapse. Integrated with auxiliary radiomic and large language model (LLM) modules, TGBP provides an end-to-end, interpretable SINS evaluation. Validated on a multi-center, multi-cancer cohort ($N=482$), TGBP achieved 90.2\% accuracy in 3-tier stability triage. In a blinded reader study ($N=30$), TGBP significantly outperformed medical oncologists on complex structural features ($κ=0.857$ vs.\ $0.570$) and prevented compounding errors in Total Score estimation ($κ=0.625$ vs.\ $0.207$), democratizing expert-level opportunistic screening.

q-bio.QM↗
arXiv2026

Curriculum Reinforcement Learning from Easy to Hard Tasks Improves LLM Reasoning

We aim to improve the reasoning capabilities of language models via reinforcement learning (RL). Recent RL post-trained models like DeepSeek-R1 have demonstrated reasoning abilities on mathematical and coding tasks. However, prior studies suggest that using RL alone to improve reasoning on inherently difficult tasks is less effective. Here, we draw inspiration from curriculum learning and propose to schedule tasks from easy to hard (E2H), allowing LLMs to build reasoning skills gradually. Our method is termed E2H Reasoner. Empirically, we observe that, although easy tasks are important initially, fading them out through appropriate scheduling is essential in preventing overfitting. Theoretically, we establish convergence guarantees for E2H Reasoner within an approximate policy iteration framework. We derive finite-sample complexity bounds and show that when tasks are appropriately decomposed and conditioned, learning through curriculum stages requires fewer total samples than direct learning. Experiments across multiple domains show that E2H Reasoner significantly improves the reasoning ability of small LLMs (1.5B to 3B), which otherwise struggle when trained with vanilla RL alone, highlighting the effectiveness of our method. Our code can be found on https://github.com/divelab/E2H-Reasoning.

cs.LG↗
arXiv2026

Two-Channel Filter Banks on Joint Time-Vertex Graphs with Oversampled Graph Laplacian Matrix

To address the limitations of conventional critically sampled graph filter banks in joint time-vertex signal processing, which require decomposing the joint graph into bipartite subgraphs and thus cannot fully exploit all temporal and spatial edges in a single-stage transform, we introduce the joint time-vertex oversampled graph Laplacian matrix. This operator enables the construction of bipartite extensions that preserve all edges of the original joint graph and supports redundant multiresolution representations. Based on this operator, we design two-channel joint time-vertex oversampled graph filter banks and develop efficient oversampling extensions using a $K$-coloring strategy. The proposed framework is applied to both graph signal and image/video denoising, modeling images as graph signals to leverage structural relationships. Extensive experiments demonstrate its effectiveness in decomposition, reconstruction, and denoising, achieving notable performance improvements over critically sampled and existing methods.

math.GM↗
arXiv2026

A Note on the Consistent-$Q$ Scheme for Odd-Odd Nuclei

Compared with even-even nuclear systems, the dynamical structure of low-lying excited states in odd-odd nuclei poses significantly greater theoretical challenges. The interacting boson fermion fermion model provides an effective framework for capturing the structural properties of odd-odd nuclei. Within this algebraic framework, the consistent-$Q$ scheme, which was widely used to map nuclear shape phase diagram, is extended to odd-odd systems to examine how unpaired nucleons influence level structures and their evolution across prototypical transitional regions. The results demonstrate that the presence of unpaired nucleons does not suppress the critical behavior characterizing level evolution across the U(5)-SU(3), U(5)-O(6) and SU(3)-O(6) shape transitions, suggesting that shape phase transitions remain a fundamental mechanism governing the low-lying structural evolution of odd-odd nuclei in the heavy and intermediately-heavy mass regions.

nucl-th↗
arXiv2026

Attention Residuals

Residual connections with PreNorm are standard in modern LLMs, yet they accumulate all layer outputs with fixed unit weights. This uniform aggregation causes uncontrolled hidden-state growth with depth, progressively diluting each layer's contribution. We propose Attention Residuals (AttnRes), which replaces this fixed accumulation with softmax attention over preceding layer outputs, allowing each layer to selectively aggregate earlier representations with learned, input-dependent weights. To address the memory and communication overhead of attending over all preceding layer outputs for large-scale model training, we introduce Block AttnRes, which partitions layers into blocks and attends over block-level representations, reducing the memory footprint while preserving most of the gains of full AttnRes. Combined with cache-based pipeline communication and a two-phase computation strategy, Block AttnRes becomes a practical drop-in replacement for standard residual connections with minimal overhead. Scaling law experiments confirm that the improvement is consistent across model sizes, and ablations validate the benefit of content-dependent depth-wise selection. We further integrate AttnRes into the Kimi Linear architecture (48B total / 3B activated parameters) and pre-train on 1.4T tokens, where AttnRes mitigates PreNorm dilution, yielding more uniform output magnitudes and gradient distribution across depth, and improves downstream performance across all evaluated tasks.

cs.CL↗
arXiv2026

HoRD: Robust Humanoid Control via History-Conditioned Reinforcement Learning and Online Distillation

Humanoid robots can suffer significant performance drops under small changes in dynamics, task specifications, or environment setup. We propose HoRD, a two-stage learning framework for robust humanoid control under domain shift. First, we train a high-performance teacher policy via history-conditioned reinforcement learning, where the policy infers latent dynamics context from recent state--action trajectories to adapt online to diverse randomized dynamics. Second, we perform online distillation to transfer the teacher's robust control capabilities into a transformer-based student policy that operates on sparse root-relative 3D joint keypoint trajectories. By combining history-conditioned adaptation with online distillation, HoRD enables a single policy to adapt zero-shot to unseen domains without per-domain retraining. Extensive experiments show HoRD outperforms strong baselines in robustness and transfer, especially under unseen domains and external perturbations. Code and project page are available at https://tonywang-0517.github.io/hord/.

cs.RO↗
arXiv2026

UAV traffic scene understanding: A regulation embedded multi-modal network and a unified benchmark

Traffic scene understanding from unmanned aerial vehicle (UAV) platforms is crucial for intelligent transportation systems due to its flexible deployment and wide-area monitoring capabilities. However, existing methods face significant challenges in real-world surveillance, as their heavy reliance on optical imagery leads to severe performance degradation under adverse illumination conditions like nighttime and fog. Furthermore, current Visual Question Answering (VQA) models are restricted to elementary perception tasks, lacking the domain-specific regulatory knowledge required to assess complex traffic behaviors. To address these limitations, we propose a novel Multi-modal Traffic Cognition Network (MTCNet) for robust UAV traffic scene understanding. Specifically, we design a Prototype-Guided Knowledge Embedding (PGKE) module that leverages high-level semantic prototypes from an external Traffic Regulation Memory (TRM) to anchor domain-specific knowledge into visual representations, enabling the model to comprehend complex behaviors and distinguish fine-grained traffic violations. Moreover, we develop a Quality-Aware Spectral Compensation (QASC) module that exploits the complementary characteristics of optical and thermal modalities to perform bidirectional context exchange, effectively compensating for degraded features to ensure robust representation in complex environments. In addition, we construct Traffic-VQA, the first large-scale optical-thermal infrared benchmark for cognitive UAV traffic understanding, comprising 8,180 aligned image pairs and 1.3 million question-answer pairs across 31 diverse types. Extensive experiments demonstrate that MTCNet significantly outperforms state-of-the-art methods in both cognition and perception scenarios. The dataset is available at https://github.com/YuZhang-2004/UAV-traffic-scene-understanding.

cs.CV↗
arXiv2026

Methods for characterization of atomic-scale field emission point-electron-source

Field emission (FE) electron sources are made close to atomic-scale to reach the highest spatial resolution as well as stable emission for electron microscopy, electron beam inspection and lithography. At present, no single agreed method exists of using FE current-voltage data to extract the apparent emission area, which is needed for predicting some beam properties. The 1956 theory of Murphy and Good (MG) is better physics than the 1920s theory of Fowler and Nordheim (FN) and colleagues, but many researchers use simplified FN theory to analyse experimental data. The present paper reports an experimental method of finding apparent emission area, based on using field ion and field electron microscopes (FIM-FEM). The discrepancy of emission area between the FIM-FEM method and MG-based analysis is a factor of 7.4, while that with simplified FN-based analysis is about 25, confirming MG theory is better for FE data analysis. The result allows deduction of key indicators, including source energy spread, reduced brightness and emission efficiency. A downloadable program is made available to help analysis. Our work provides a new experimental method of characterizing FE electron sources, especially the atomic-scale cold cathode, for which existing plot-based data-analysis methods are not suitable.

quant-ph↗
arXiv2026

Bridging Sequential and Contextual Features with a Dual-View of Fine-grained Core-Behaviors and Global Interest-Distribution

Click-through rate (CTR) prediction tasks typically estimate the probability of a user clicking on a candidate item by modeling both user behavior sequence features and the item's contextual features, where the user behavior sequence is particularly critical as it dynamically reflects real-time shifts in user interest. Traditional CTR models often aggregate this dynamic sequence into a single vector before interacting it with contextual features. This approach, however, not only leads to behavior information loss during aggregation but also severely limits the model's capacity to capture interactions between contextual features and specific user behaviors, ultimately impairing its ability to capture fine-grained behavioral details and hindering models' prediction accuracy. Conversely, a naive approach of directly interacting with each user action with contextual features is computationally expensive and introduces significant noise from behaviors irrelevant to the candidate item. This noise tends to overwhelm the valuable signals arising from interactions involving more behaviors relevant to the candidate item. Therefore, to resolve the above issue, we propose a Core-Behaviors and Distributional-Compensation Dual-View Interaction Network (CDNet), which bridges the gap between sequential and contextual feature interactions from two complementary angles: a fine-grained interaction involving the most relevant behaviors and contextual features, and a coarse-grained interaction that models the user's overall interest distribution against the contextual features. By simultaneously capturing important behavioral details without forgoing the holistic user interest, CDNet effectively models the interplay between sequential and contextual features without imposing a significant computational burden. Ultimately, extensive experiments validate the effectiveness of CDNet.

cs.IR↗
arXiv2026

Deferred is Better: A Framework for Multi-Granularity Deferred Interaction of Heterogeneous Features

Click-through rate (CTR) prediction models estimates the probability of a user-item click by modeling interactions across a vast feature space. A fundamental yet often overlooked challenge is the inherent heterogeneity of these features: their sparsity and information content vary dramatically. For instance, categorical features like item IDs are extremely sparse, whereas numerical features like item price are relatively dense. Prevailing CTR models have largely ignored this heterogeneity, employing a uniform feature interaction strategy that inputs all features into the interaction layers simultaneously. This approach is suboptimal, as the premature introduction of low-information features can inject significant noise and mask the signals from information-rich features, which leads to model collapse and hinders the learning of robust representations. To address the above challenge, we propose a Multi-Granularity Information-Aware Deferred Interaction Network (MGDIN), which adaptively defers the introduction of features into the feature interaction process. MGDIN's core mechanism operates in two stages: First, it employs a multi-granularity feature grouping strategy to partition the raw features into distinct groups with more homogeneous information density in different granularities, thereby mitigating the effects of extreme individual feature sparsity and enabling the model to capture feature interactions from diverse perspectives. Second, a delayed interaction mechanism is implemented through a hierarchical masking strategy, which governs when and how each group participates by masking low-information groups in the early layers and progressively unmasking them as the network deepens. This deferred introduction allows the model to establish a robust understanding based on high-information features before gradually incorporating sparser information from other groups...

cs.IR↗
arXiv2026

Taming the Long Tail: Denoising Collaborative Information for Robust Semantic ID Generation

Item IDs form the backbone of industrial recommender systems, but suffer from representation instability and poor long-tail generalization in large, dynamic item corpora. Semantic IDs (SIDs) mitigate these issues by enabling knowledge sharing through quantization of item content features. Existing methods attempt to enhance SID expressiveness by incorporating collaborative information with content features; however, they often overlook a critical distinction: unlike relatively uniform content features, user-item interactions are highly skewed, resulting in a significant quality gap in collaborative information between popular and long-tail items. This mismatch leads to two critical limitations: (1) Collaborative Noise Corrupts Behavior-Content Alignment: Behavior-content alignment is a prevailing approach for modeling shared information. However, indiscriminate alignment allows collaborative noise from long-tail items to corrupt their content representations, leading to the loss of critical multimodal information. (2) Collaborative Noise Obscures Critical Behavioral SIDs: When modeling modality-specific information, prior works typically generate multiple behavioral SIDs with equal weights for each item. This equal-weight scheme fails to reflect the varying importance of different behavioral SIDs, making it difficult for downstream tasks to distinguish informative SIDs from noisy ones. To address these challenges, we propose ADC-SID, a framework that Adaptively Denoises Collaborative information for SID quantization. It comprises two key components: (i) Adaptive Behavior-Content Alignment, which adjusts alignment strength to mitigate corruption caused by collaborative noise; and (ii) Dynamic Behavioral Weighting Mechanism, which learns importance scores for behavioral SIDs to enable downstream models to suppress noise. Extensive experiments has demonstrated ADC-SID's superiority...

cs.IR↗
arXiv2026

PLANING: A Loosely Coupled Triangle-Gaussian Framework for Streaming 3D Reconstruction

Streaming reconstruction from monocular image sequences remains challenging, as existing methods typically favor either high-quality rendering or accurate geometry, but rarely both. We present PLANING, an efficient on-the-fly reconstruction framework built on a hybrid representation that loosely couples explicit geometric primitives with neural Gaussians, enabling geometry and appearance to be modeled in a decoupled manner. This decoupling supports an online initialization and optimization strategy that separates geometry and appearance updates, yielding stable streaming reconstruction with substantially reduced structural redundancy. PLANING improves dense mesh Chamfer-L2 by 18.52% over PGSR, surpasses ARTDECO by 1.31 dB PSNR, and reconstructs ScanNetV2 scenes in under 100 seconds, over 5x faster than 2D Gaussian Splatting, while matching the quality of offline per-scene optimization. Beyond reconstruction quality, the structural clarity and computational efficiency of PLANING make it well suited for a broad range of downstream applications, such as enabling large-scale scene modeling and simulation-ready environments for embodied AI. Project page: https://city-super.github.io/PLANING/ .

cs.CV↗
arXiv2026

Long-range magnetic order with disordered spin orientations in a high-entropy antiferromagnet

Disorder in magnetic systems typically suppresses long-range order, promoting short-range states such as spin glasses and magnetic clusters. This is particularly prominent in high-entropy materials, characterized by the random distributions of local magnetic entities and exchange interactions. However, in rare exceptions, long-range magnetic order can persist in high-entropy systems, while the microscopic characters and underlying mechanisms remain elusive, especially the magnetic behaviors of individual elements. Here, combining neutron diffraction and resonant soft x-ray scattering, we have conducted an element-specific investigation into the magnetic order of a high-entropy honeycomb-lattice van der Waals material (Mn1/4Fe1/4Co1/4Ni1/4)PS3. Despite significant atomic disorder, long-range zigzag antiferromagnetic order is observed below 72 K, with all four transition-metal elements participating in a unified phase transition. However, the spin orientations of various elements are distinct, attributed to the competition between single-ion anisotropies and exchange interactions. Our findings showcase a novel form of long-range magnetic order with disordered spin orientations, which is synergically stabilized by distinct magnetic elements in a high entropy magnet, offering a new paradigm for understanding complex magnetic systems.

cond-mat.str-el↗
arXiv2026

Adaptive Clinical-Aware Latent Diffusion for Multimodal Brain Image Generation and Missing Modality Imputation

Multimodal neuroimaging provides complementary insights for Alzheimer's disease diagnosis, yet clinical datasets frequently suffer from missing modalities. We propose ACADiff, a framework that synthesizes missing brain imaging modalities through adaptive clinical-aware diffusion. ACADiff learns mappings between incomplete multimodal observations and target modalities by progressively denoising latent representations while attending to available imaging data and clinical metadata. The framework employs adaptive fusion that dynamically reconfigures based on input availability, coupled with semantic clinical guidance via GPT-4o-encoded prompts. Three specialized generators enable bidirectional synthesis among sMRI, FDG-PET, and AV45-PET. Evaluated on ADNI subjects, ACADiff achieves superior generation quality and maintains robust diagnostic performance even under extreme 80\% missing scenarios, outperforming all existing baselines. To promote reproducibility, code is available at https://github.com/rongzhou7/ACADiff

cs.CV↗