arXiv subjects

Yuhan Wang

Publications and source records attributed to Yuhan Wang.

At least 73 records · Page 4Linked to original sources

arXiv2026

Training-Free Rate-Distortion-Perception Traversal With Diffusion

The rate-distortion-perception (RDP) tradeoff characterizes the fundamental limits of lossy compression by jointly considering bitrate, reconstruction fidelity, and perceptual quality. While recent neural compression methods have improved perceptual performance, they typically operate at fixed points on the RDP surface, requiring retraining to target different tradeoffs. In this work, we propose a training-free framework that leverages pre-trained diffusion models to traverse the entire RDP surface. Our approach integrates a reverse channel coding (RCC) module with a novel score-scaled probability flow ODE decoder. We theoretically prove that the proposed diffusion decoder is optimal for the distortion-perception tradeoff under AWGN observations and that the overall framework with the RCC module achieves the optimal RDP function in the Gaussian case. Empirical results across multiple datasets demonstrate the framework's flexibility and effectiveness in navigating the ternary RDP tradeoff using pre-trained diffusion models. Our results establish a practical and theoretically grounded approach to adaptive, perception-aware compression.

cs.IT↗

arXiv2026

S2Aligner: Pair-Efficient and Transferable Pre-Training for Sparse Text-Attributed Graphs

Pre-training on text-attributed graphs (TAGs) is central to building transferable graph foundation models, where LLM-as-Aligner methods align graph and text representations through the semantic knowledge of large language models. However, these methods usually assume that node texts provide sufficient and reliable supervision, an assumption often violated in real-world sparse TAGs. When textual anchors are missing, noisy, or uneven across domains, graph structures must be aligned with weak semantic evidence, leading to unreliable structure-semantics correspondence and sparsity-induced transfer bias. This paper presents S2Aligner, a sparsity-aware and structure-enhanced LLM-as-Aligner framework for graph-text pre-training on sparse TAGs. The key idea is to decouple semantic alignment from structural modeling, allowing topology-aware signals to enhance alignment without contaminating the shared semantic space. Specifically, S2Aligner decomposes graph-text representations into semantic and structural components, uses structure-oriented reconstruction with consistency control to inject reliable topology cues into text representations, and suppresses inconsistent structural signals under textual sparsity. Moreover, S2Aligner introduces sparsity-aware cross-domain risk balancing, which calibrates domain risks through a global-domain density ratio and downweights unreliable sparse samples via graph reliability estimation. Theoretical analysis shows that this objective reduces cross-domain generalization gaps by controlling domain risk discrepancy. Extensive experiments across diverse graph domains, sparsity levels, and downstream tasks demonstrate that S2Aligner consistently outperforms existing baselines.

cs.LG↗

arXiv2026

Divergence Meets Consensus: A Multi-Source Negative Sampling Framework for Sequential Recommendation

Negative sampling is significant for training sequential recommendation models under implicit feedback. The predominant strategy, self-guided hard negative sampling, selects negatives based on the model's current state but suffers from three limitations: (1) the coupling between sampling and model updates triggers a vicious cycle that drives the model into local optima; (2) relying on current model parameters narrows sampling to a small region of the item space, reducing diversity and harming generalization; (3) identifying a hard negative requires scoring the entire candidate pool, causing substantial computational overhead with minimal information gain. To address these challenges, we propose MDCNS (Multi-source Divergence-Consensus for Negative Sampling), a novel "Teacher-Peer-Self" framework inspired by Vygotsky's Zone of Proximal Development (ZPD) theory. The proposed method comprises three components, including multi-source scoring, divergence re-ranking, and consensus distillation. Firstly, multi-source scoring incorporates peer and ensemble teacher models to inject external negative signals and break the self-reinforcement loop. Then, divergence re-ranking exploits prediction discrepancy between self and peer models to enhance sampling diversity. Finally, consensus distillation aligns the self model with the teacher via KL divergence, simultaneously improving computational cost utilization. Extensive experiments on six real-world datasets and five backbone models show that MDCNS consistently outperforms state-of-the-art negative sampling methods, demonstrating strong effectiveness and generalization.

cs.IR↗

arXiv2026

Graph-Based Financial Fraud Detection with Calibrated Risk Scoring and Structural Regularization

Financial transaction fraud prevention faces challenges such as complex relationship structures, concealed behavioral patterns, and dynamically changing data distribution. Discrimination models relying solely on independent sample features are insufficient to fully characterize the risks of group collaboration and chain transfers within transaction networks. This paper proposes a graph neural network representation learning and risk discrimination framework for financial transaction fraud prevention. It integrates transaction records and identity information into node attributes and constructs a transaction graph based on shared attributes and interaction consistency to explicitly model inter-transaction relationships. In model design, a multi-layer message passing mechanism is employed to aggregate neighborhood information, learn node embedding representations containing structural context semantics, and output transaction-level fraud probability and risk scores through a lightweight risk discrimination head. A weighted supervision objective is introduced to mitigate training bias caused by class imbalance, and structural consistency regularization constraints are combined to suppress the impact of noisy edges on representation drift, thereby improving the stability and usability of risk characterization. Experiments are conducted on a publicly available financial transaction dataset, comparing various methods in the same direction and comprehensively evaluating them under a unified evaluation protocol. The results show that the proposed method outperforms other methods in risk ranking and probability calibration quality, validating the effectiveness of graph structure modeling and representation learning collaboration in financial transaction fraud prevention.

cs.LG↗

arXiv2026

SDiaReward: Modeling and Benchmarking Spoken Dialogue Rewards with Modality and Colloquialness

The rapid evolution of end-to-end spoken dialogue systems demands transcending mere textual semantics to incorporate paralinguistic nuances and the spontaneous nature of human conversation. However, current methods struggle with two critical gaps: the modality gap, involving prosody and emotion, and the colloquialness gap, distinguishing written scripts from natural speech. To address these challenges, we introduce SDiaReward, an end-to-end multi-turn reward model trained on SDiaReward-Dataset, a novel collection of episode-level preference pairs explicitly targeting these gaps. It operates directly on full multi-turn speech episodes and is optimized with pairwise preference supervision, enabling joint assessment of modality and colloquialness in a single evaluator. We further establish ESDR-Bench, a stratified benchmark for robust episode-level evaluation. Experiments demonstrate that SDiaReward achieves state-of-the-art pairwise preference accuracy, significantly outperforming general-purpose audio LLMs. Further analysis suggests that SDiaReward captures relative conversational expressiveness beyond superficial synthesis cues, improving generalization across domains and recording conditions. Code, data, and demos are available at https://github.com/MM-Speech/SDiaReward/.

eess.AS↗

arXiv2026

Any2Any 3D Diffusion Models with Knowledge Transfer: A Radiotherapy Planning Study

Voxel-wise dose prediction is a critical yet challenging task in practical radiotherapy (RT) planning, as bespoke models trained from scratch often struggle to generalize across diverse clinical settings. Meanwhile, generative models trained on billion-scale datasets from vision domains have achieved impressive performance. Herein, we propose DiffKT3D, a unified Any2Any 3D diffusion framework that leverages prior knowledge from pretrained video diffusion models for efficient and clinically meaningful dose prediction. To enable flexible conditioning across multiple clinical modalities (CT, anatomical structures, body, beam settings, etc.), we introduce an Any2Any conditional paradigm utilizing modality-specific embeddings without cross-attention overhead. Further, we design a novel reinforcement learning (RL) post-training mechanism guided by a clinically-informed Scorecard explicitly tailored to institutional treatment preferences. Compared with winner of GDP-HMM challenge, DiffKT3D sets a new state-of-the-art in dose prediction by reducing voxel-level MAE from 2.07 to 1.93. In addition, DiffKT3D achieves superior image quality and preference match. These results demonstrate that transferring diffusion priors via modality-aware conditioning and clinically aligned RL post-training can provide a robust and generalizable solution for RT planning across various clinical scenarios.

cs.CV↗

arXiv2026

Unsupervised Graph Modeling for Anomaly Detection in Accounting Subject Relationships

This paper addresses the problem of anomaly detection in accounting subject association structures, proposing a structured modeling and unsupervised discriminant framework based on graph neural networks. This framework is used to mine stable correspondences between subjects and identify structural deviations from general ledger details and voucher entries. The method first abstracts accounting subjects as graph nodes, and the co-occurrence and debit/credit correspondence of subjects in the same business record are abstracted as weighted edges. The edge weights are characterized by statistical measures such as co-occurrence frequency or amount aggregation, thus forming a period-level accounting subject association graph. In the representation learning stage, a message passing mechanism is used to fuse the node's own attributes and neighborhood context to obtain node embeddings containing structural information. In the anomaly detection stage, the rationality of subject pair connections is estimated through a relation reconstruction decoder, and edge-level anomaly scores are defined based on the degree of deviation in reconstruction probabilities. These scores are then aggregated to obtain node-level risk ranking and local anomaly localization. This framework can simultaneously capture local substructure anomalies and cross-community anomaly connections without relying on anomaly labeling, outputting traceable subject pair risk clues. Comparative experiments demonstrate more stable comprehensive discriminant capabilities and higher top-ranking accuracy.

cs.LG↗

arXiv2026

Probing phonon chirality and circular lattice motion with symmetry-selective nonlinear optical spectroscopy

Truly chiral phonons are lattice eigenmodes that combine broken mirror symmetry with circular atomic motion. They can mediate angular-momentum-selective interactions in quantum materials, yet directly resolving both their chirality and underlying circular motion remains challenging, especially in high-symmetry crystals. Here we show that symmetry-selective terahertz difference-frequency spectroscopy provides a phase- and polarization-resolved route to identifying truly chiral phonons in a tabletop experiment. Using $α$-quartz as a benchmark, we validate this approach by resolving phonon chirality via chiral-sensitive $χ^{(2)}_{ijk}$ tensor elements ($i \neq j \neq k$), while vector-field detection directly reveals a time-dependent polarization rotation arising from circular ionic motion and thus nonzero angular momentum. Applying the same protocol to tetragonal $α$-TeO$_2$, we isolate chiral $E$-mode resonances below 5~THz and directly verify their circular lattice motion, thereby resolving a symmetry-imposed ambiguity in chiral-phonon identification in fourfold-symmetric crystals. Our results establish symmetry-selective nonlinear terahertz spectroscopy as a general route to identify truly chiral phonons in condensed matter systems.

physics.optics↗

arXiv2026

PRL-Bench: A Comprehensive Benchmark Evaluating LLMs' Capabilities in Frontier Physics Research

The paradigm of agentic science requires AI systems to conduct robust reasoning and engage in long-horizon, autonomous exploration. However, current scientific benchmarks remain confined to domain knowledge comprehension and complex reasoning, failing to evaluate the exploratory nature and procedural complexity of real-world research. In this work, we present research-oriented evaluations in theoretical and computational physics, a natural testbed with comprehensive domain knowledge, complex reasoning, and verifiable end-to-end workflows without reliance on experiments. Here we introduce PRL-Bench (Physics Research by LLMs), a benchmark designed to systematically map the capability boundaries of LLMs in executing end-to-end physics research. Constructed from 100 curated papers from the latest issues of Physical Review Letters since August 2025 and validated by domain experts, PRL-Bench covers five major theory- and computation-intensive subfields of modern physics: astrophysics, condensed matter physics, high-energy physics, quantum information, and statistical physics. Each task in the benchmark is designed to replicate the core properties of authentic scientific research, including exploration-oriented formulation, long-horizon workflows, and objective verifiability, thereby reconstructing the essential reasoning processes and research workflows of real physics research. Evaluation across frontier models shows that performance remains limited, with the best overall score below 50, revealing a pronounced gap between current LLM capabilities and the demands of real scientific research. PRL-Bench serves a reliable testbed for accessing next generation AI scientists advancing AI systems toward autonomous scientific discovery.

cs.LG↗

arXiv2026

Optimization of Laser Irradiation Uniformity for the Double-Cone Ignition Scheme with MULTI-3D simulations

The double-cone ignition (DCI) scheme holds a promising perspective for laser driven fusion energy and astrophysics. However, optimizing the laser irradiation uniformity under the constraints of limited laser beams and a given cone angle remains to be explored. We utilized the three-dimensional radiation hydrodynamics program MULTI-3D to simulate the interaction process between the laser and plasma shell. By employing Bayesian optimization for the pointing position of the incident laser beams, we achieved a laser irradiation scheme with nonuniformity less than 5%. This study can provide references for experiments and offer valuable insights for other laser fusion schemes.

physics.plasm-ph↗

arXiv2026

S-wave kaon condensation in neutron-star matter within a chiral model framework with dynamical meson masses

We investigate s-wave kaon condensation in dense matter and neutron stars within the updated Chiral Mean Field model with an improved meson description (mCMF), which incorporates dynamically generated in-medium meson masses arising from explicit chiral symmetry breaking and vector-meson self-interactions. In contrast to conventional relativistic mean-field descriptions with constant meson masses, the mCMF framework introduces a self-consistent feedback between the meson sector and the dense-matter equations of motion. The kaon dispersion relation is derived from the nonlinear SU(3) Lagrangian, including the Weinberg-Tomozawa interaction and additional baryon-pseudoscalar couplings, and the onset of condensation is determined under conditions of charge neutrality and $β$ equilibrium. Our calculations include the full baryon octet together with electrons and muons at zero temperature. We analyze the impact of hyperons, muons, and kaon condensation on the equation of state, on neutron-star mass--radius relations, and neutron-star thermal evolution, and examine the sensitivity of the onset density and stellar properties to variations in the nucleon--kaon scattering length and to different model vector parameters and vector self-interactions. We find that $K^{-}$ condensation sets in between $n \sim (2-8)\, n_0$ (in units of nuclear saturation density) and leads to a moderate to strong softening (in one case, a slight stiffening of the equation of state), depending on the interplay of kaons and hyperons, while remaining compatible with current $2\,M_\odot$ and small-radius neutron-star observational constraints and producing distinguishable behavior in the neutron-star cooling. This work provides an improved and thermodynamically consistent framework for studying exotic degrees of freedom in neutron-star matter.

nucl-th↗

arXiv2026

Kestrel: Grounding Self-Refinement for LVLM Hallucination Mitigation

Large vision-language models (LVLMs) have become increasingly strong but remain prone to hallucinations in multimodal tasks, which significantly narrows their deployment. As training these LVLMs to avoid hallucinations becomes prohibitively expensive for larger models, training-free methods offer a cheap and flexible solution to this problem, yet existing approaches based on decoding or tool use often bring limited gains and/or weak interpretability. We propose Kestrel, a training-free framework for LVLM hallucination mitigation that combines an explicit visual-grounding agent with evidence-verified self-refinement mechanism. In detail, Kestrel first collects explicit visual evidence and converts tool outputs into reusable and structured textual evidence. Second, to take full advantage of these evidence, Kestrel verifies them via an LVLM judge for evidence checking, then iteratively self-refine answers based on verified evidence to reduce the risk of over-correction. Extensive experiments show that Kestrel improves performance over strong baselines across hallucination benchmarks (e.g., average +3.31% on POPE and +28.34 on MME-Hallucination with Qwen3-VL), while providing transparent verification traces for hallucination diagnosis and analysis -- e.g., both the integrated self-refinement module and grounding agent contributing an average +2.0% gain on POPE.

cs.CV↗

arXiv2026

CCAT: Mod-Cam Readout Overview and Flexible Stripline Performance

The CCAT Observatory's primary science instrument, Prime-Cam, is nearing readiness for deployment to the Fred Young Submillimeter Telescope (FYST) in the Atacama Desert in northern Chile. When fully deployed, Prime-Cam will field approximately 100,000 kinetic inductance detectors (KIDs) across seven instrument modules making both broadband and polarimetric measurements. Meanwhile, in-lab characterization of the first CCAT instrument module, a 280 GHz broadband camera fielding over 10,000 KIDs, is currently underway in the testbed instrument Mod-Cam. Both Mod-Cam and Prime-Cam will employ 46 cm long low-thermal-conductivity flexible circuits ("stripline") between 4 K and 300 K to connect large-format arrays of multiplexed KIDs in each instrument module to readout electronics. The 280 GHz camera currently installed in Mod-Cam uses six of these striplines to read out its over 10,000 detectors. We have examined the thermal and electrical performance of the stripline installed in Mod-Cam. We begin by characterizing the OFHC copper in the stripline traces, allowing for the estimation of thermal loading through these flexible circuits in their configurations in both Mod-Cam and Prime-Cam. We then directly measure the thermal conductivity of the stripline, finding it is best described by $kA = 22\pm6~T^{0.84\pm0.09}~\mathrm{μ~W~m~K^{-1}}$ for temperature ranges of 6 K < T < 20 K and $kA~=~0.6\pm0.3~T^{-0.4\pm0.1}~\mathrm{mW~m~K^{-1}}$ for ranges from 20 K < T < 80 K. Following our thermal characterizations, we report on the transmission and crosstalk properties of the Mod-Cam readout chain, isolating elevated crosstalk to SMP-SMA transition printed circuit boards (PCBs) that interface with the stripline. This finding validates the stripline circuit as a viable high-density cabling option for large-format array readout.

astro-ph.IM↗

arXiv2026

CCAT: Magnetic Sensitivity Measurements of Kinetic Inductance Detectors

The CCAT Observatory is a ground-based submillimeter to millimeter experiment located on Cerro Chajnantor in the Atacama Desert, at an altitude of 5,600 meters. CCAT features the 6-meter Fred Young Submillimeter Telescope (FYST), which will cover frequency bands from 210 GHz to 850 GHz using its first-generation science instrument, Prime-Cam. The detectors used in Prime-Cam are feedhorn-coupled, lumped-element superconducting microwave kinetic inductance detectors (KIDs). The telescope will perform wide-area surveys at speeds on the order of degrees per second. During telescope operation, the KIDs are exposed to changes in the magnetic field caused by the telescope's movement through Earth's magnetic field and internal sources within the telescope. We present and compare measurements of the magnetic sensitivity of three different CCAT KID designs at 100 mK. The measurements are conducted in a dilution refrigerator (DR) with a set of room temperature Helmholtz coils positioned around the DR. We discuss the implications of these results for CCAT field operations.

astro-ph.IM↗

arXiv2026

The Simons Observatory: Characterization of the 220/280 GHz TES Detector Modules

The Simons Observatory (SO) is a new suite of cosmic microwave background telescopes in the Chilean Atacama Desert with an extensive science program spanning cosmology, Galactic and extragalactic astrophysics, and particle physics. SO will survey the millimeter-wave sky over a wide range of angular scales using six spectral bands across three types of dichroic, polarization-sensitive transition-edge sensor (TES) detector modules: Low-Frequency (LF) modules with bandpasses centered near 30 and 40 GHz, Mid-Frequency (MF) modules near 90 and 150 GHz, and Ultra-High-Frequency (UHF) modules near 220 and 280 GHz. Twenty-five UHF detector modules, each containing 1720 optically-coupled TESs connected to microwave SQUID multiplexing readout, have now been produced. This work summarizes the pre-deployment characterization of these detector modules in laboratory cryostats. Across all UHF modules, we find an average operable TES yield of 83%, equating to over 36,000 devices tested. The distributions of (220, 280) GHz saturation powers have medians of (24, 26) pW, near the centers of their target ranges. For both bands, the median optical efficiency is 0.6, the median effective time constant is 0.4 ms, and the median dark noise-equivalent power (NEP) is ~40 aW/rtHz. The expected photon NEPs at (220, 280) GHz are (64, 99) aW/rtHz, indicating these detectors will achieve background-limited performance on the sky. Thirty-nine UHF and MF detector modules are currently operating in fielded SO instruments, which are transitioning from the commissioning stage to full science observations.

astro-ph.IM↗

arXiv2026

RIS-Aided Wireless Amodal Sensing for Single-View 3D Reconstruction

Amodal sensing is critical for various real-world sensing applications because it can recover the complete shapes of partially occluded objects in complex environments. Among various amodal sensing paradigms, wireless amodal sensing is a potential solution due to its advantages of environmental robustness, privacy preservation, and low cost. However, the sensing data obtained by wireless system is sparse for shape reconstruction because of the low spatial resolution, and this issue is further intensified in complex environments with occlusion. To address this issue, we propose a Reconfigurable Intelligent Surface (RIS)-aided wireless amodal sensing scheme that leverages a large-scale RIS to enhance the spatial resolution and create reflection paths that can bypass the obstacles. A generative learning model is also employed to reconstruct the complete shape based on the sensing data captured from the viewpoint of the RIS. In such a system, it is challenging to optimize the RIS phase shifts because the relationship between RIS phase shifts and amodal sensing accuracy is complex and the closed-form expression is unknown. To tackle this challenge, we develop an error prediction model that learns the mapping from RIS phase shifts to amodal sensing accuracy, and optimizes RIS phase shifts based on this mapping. Experimental results on the benchmark dataset show that our method achieves at least a 56.73% reduction in reconstruction error compared to conventional schemes under the same number of RIS configurations.

eess.SP↗

arXiv2026

The Simons Observatory: On-sky performance of radio-transparent multi-layer insulation (RT-MLI) using Styroace-II Styrofoam

We present the on-sky performance of a Radio-Transparent Multi-Layer Insulation filter (RT-MLI) that uses Styroace-II styrofoam to reject ambient thermal radiation from entering a 0.42 m diameter aperture to a sub-100 mK bolometric detector array cooled by a dilution-refrigerator. We find that greater than 90% of the expected incident infra-red (IR) radiation is rejected, resulting in $<$12 W of measured transmitted power. Transmitted power in the detector passbands is consistent with a lower bound of 95%. We address filter design and placement, thermal loading, and mm-wave transmission.

astro-ph.IM↗

arXiv2025

UltraShape 1.0: High-Fidelity 3D Shape Generation via Scalable Geometric Refinement

In this report, we introduce UltraShape 1.0, a scalable 3D diffusion framework for high-fidelity 3D geometry generation. The proposed approach adopts a two-stage generation pipeline: a coarse global structure is first synthesized and then refined to produce detailed, high-quality geometry. To support reliable 3D generation, we develop a comprehensive data processing pipeline that includes a novel watertight processing method and high-quality data filtering. This pipeline improves the geometric quality of publicly available 3D datasets by removing low-quality samples, filling holes, and thickening thin structures, while preserving fine-grained geometric details. To enable fine-grained geometry refinement, we decouple spatial localization from geometric detail synthesis in the diffusion process. We achieve this by performing voxel-based refinement at fixed spatial locations, where voxel queries derived from coarse geometry provide explicit positional anchors encoded via RoPE, allowing the diffusion model to focus on synthesizing local geometric details within a reduced, structured solution space. Our model is trained exclusively on publicly available 3D datasets, achieving strong geometric quality despite limited training resources. Extensive evaluations demonstrate that UltraShape 1.0 performs competitively with existing open-source methods in both data processing quality and geometry generation. All code and trained models will be released to support future research.

cs.CV↗