arXiv subjects

Yang Liu

Publications and source records attributed to Yang Liu.

At least 163 records · Page 9Linked to original sources

arXiv2026

ConPro: Contrast Projection Pretraining for Label-Efficient Vessel Segmentation in DSA Sequences

Dense vessel annotation in digital subtraction angiography (DSA) is labor-intensive, yet every unlabeled sequence records how contrast passes through the vessels. Semi-supervised methods take their targets from the current model, and generic self-supervised pretexts reconstruct static appearance, so this signal goes unused. We propose ConPro, a self-supervised pretraining scheme whose target is a contrast projection, the normalized drop of every pixel below its temporal median over the sequence. On DIAS and DSCA, with 10%, 20% and 50% of the training cases labeled, ConPro improves on training from scratch at every label fraction and is the best of the compared methods on DSCA at 20% and 50% labels. Controlled comparisons show that the gain comes from the target. A temporal-median target with the same input, loss and budget stays at scratch level, and using the projection directly instead of learning it, as an input channel or a pseudo-label, helps little or hurts. ConPro provides pretrained weights without changing the segmentation architecture, so it combines with semi-supervised training, and UniMatch, the strongest baseline, gains 0.5 to 2.0 Dice and 0.9 to 2.3 clDice at every label fraction when started from ConPro weights, reaching 75.4 Dice on DIAS and 81.3 on DSCA.

cs.CV↗

arXiv2026

T-RoPE: Time-Aware Rotary Position Embedding for Sequential Recommendation

Large-scale recommenders increasingly adopt the sequential generative recipe behind large language models, bringing the Transformer into recommendation along with design choices made for text, including Rotary Position Embedding (RoPE). In language models, RoPE encodes token indices for relative position reasoning, but in recommendation, an interaction index records only event order, saying nothing about elapsed time, behavioral cycles across scales, or calendar phase. We revisit this choice and propose T-RoPE, a time-aware RoPE for sequential generative recommendation that replaces index-only rotation with timestamp-based angles, learnable temporal coefficients, multiscale frequency banks, shifted query alignment, and non-stationary key rotation. We prove that standard RoPE, even on timestamps, remains time-translation invariant and cannot distinguish seasonal contexts, and that T-RoPE breaks this invariance while preserving the RoPE interface. Across five public benchmarks, T-RoPE achieves the best result on every metric on every dataset, improving over the strongest baseline by 78--130\% in HR@10 on the sparse PixelRec data and 8--12\% across metrics on Amazon Books. On an industrial-scale e-commerce dataset with more than 6B interactions, it improves every metric over the HSTU + Time RAB backbone by 13--82\%, with ablations attributing the largest gains to multiscale frequencies ($+56\%$ NDCG@50) and non-stationary keys ($+4\%$). An online A/B test in the Shop app yields positive lifts in conversion rate ($+0.33\%$) and order count ($+0.63\%$). We also provide forward and backward algorithms whose added cost is linear in sequence length and head dimension, keeping time-aware RoPE practical for large generative recommenders.

cs.AI↗

arXiv2026

UR$^2$: Unify RAG and Reasoning through Reinforcement Learning

Large Language Models (LLMs) have shown strong capabilities through two complementary paradigms: Retrieval-Augmented Generation (RAG) for knowledge grounding and Reinforcement Learning from Verifiable Rewards (RLVR) for complex reasoning. However, existing attempts to unify these paradigms remain narrow in scope, typically limited to open-domain QA with fixed retrieval settings, which constrains generalization to broader domains. To address this limitation, we propose UR$^2$ (Unified RAG and Reasoning)), a general reinforcement learning framework that dynamically coordinates retrieval and reasoning. UR$^2$ introduces two key designs: a difficulty-aware curriculum that selectively invokes retrieval only for challenging instances, and a hybrid knowledge access strategy that combines domain-specific offline corpora with on-the-fly LLM-generated summaries. Together, these components mitigate the imbalance between retrieval and reasoning and improve robustness to noisy information. Experiments on open-domain QA, MMLU-Pro, medical, and mathematical reasoning tasks show that UR$^2$, built on Qwen-2.5-3/7B and LLaMA-3.1-8B, consistently outperforms existing RAG and RL baselines, and achieves performance comparable to GPT-4o-mini and GPT-4.1-mini on several benchmarks. Code, models, and data are available at https://github.com/Tsinghua-dhy/UR2.

cs.CL↗

arXiv2026

Safety Does Not Compose: Non-Decaying Loop State for Autonomous LLM Agents

Large language model agents are increasingly deployed as autonomous loops. Starting from one human goal, such a system repeatedly discovers work, plans, executes tool calls, verifies outcomes and persists state across many unattended iterations. The agent safeguards in wide use, however, are defined over a single trajectory, and their safety state is re-initialized when the next trajectory begins. We show that this is a failure of composition rather than an implementation detail. Our central result is a separation: against an attack whose evidence is fragmented across several iterations, every trajectory-scoped monitor has a true-positive rate equal to its false-positive rate, however expressive it is, because the evidence it would need never appears in the window it sees, whereas a monitor retaining cross-iteration state separates the two perfectly. We further show that the obvious repair of carrying a geometrically decaying risk score is insufficient, because the cooling-off period a patient adversary must wait is a constant that does not grow with the horizon $N$. We then present LoopHarness, which restores a persistent, non-decaying safety state at the loop level. Under mediated commits and an arbiter detection floor $δ_M$, it bounds the expected number of unauthorized irreversible actions by $B+m-1+m/δ_M$, a constant in $N$, of which the $B+m-1$ term is decided by a model-free rule and therefore survives a fully colluding verifier. We give a complete evaluation protocol on native Agent-SafetyBench tasks with paired clean and attacked episodes, an outer-state attack suite whose decisive evidence exists only across iterations, per-module ablations, and an adaptive white-box red team.

cs.CR↗

arXiv2026

SabreAgent: Language Models at Design Time for Lost-Sales Inventory Control

SabreAgent uses a language model at design time to construct two components for lost-sales inventory control: a product-specific seasonal prior and a validation-selected capped base-stock policy family. During operation, statistical forecasting and inventory optimization use these frozen artifacts to determine orders, with zero language-model calls. We evaluate the approach on the $1{,}320$ instances of InventoryBench. Under the benchmark's cost assumptions, the operations-research core draws on a zero-lead-time optimality result and a projected-inventory rule for positive deterministic lead times. The latter computes replenishment shortfalls by propagating inventory using sales along simulated demand paths. The seasonal prior adds forecast variants alongside the original forecaster, and the selected policy family handles stochastic lead times with order destruction. SabreAgent scores $0.6311$, compared with $0.5380$ for the strongest published baseline, and ranks first in all six benchmark cells. Ablations attribute most of the gain to the OR core. In the paired analysis, the seasonal component adds $1.79\%$ across the three real-data cells, and the search component adds $2.3\%$ across the two stochastic-lead-time cells. These results demonstrate how model-generated priors and policy structure can improve an OR controller through design-time use.

cs.CE↗

arXiv2026

Recasting the Destroy Step of Large Neighborhood Search as Dense Segmentation

Large neighborhood search improves an incumbent by releasing selected variables and repairing the resulting subproblem under a time limit. FOVEA casts variable selection as dense semantic segmentation on a fixed $128\times128$ canvas. A small U-Net reads twelve semantic channels and predicts cell scores, which are decoded into a variable set subject to a fixed variable budget. Family-specific layouts connect the search state to the canvas, while one set of network weights serves four problem families. The FP32 network input occupies $786$\,kB across instance sizes. Our analysis bounds constraint coupling for spatially coherent regions, gives a rank certificate for rounds with zero possible improvement, and bounds the coupling advantage attainable on an expander family. On the tested families and hardware, FOVEA trails the strongest graph encoder at $10^{4}$ variables and overtakes it near $1.5\times10^{4}$; the cost model provides an approximate crossover estimate. At larger sizes, where the graph-encoder baselines exceed device memory, FOVEA reduces the primal integral by $12$ to $16\%$ relative to the best runnable baselines. On the expander family, FOVEA performs comparably to random selection, with a coupling advantage close to one.

eess.IV↗

arXiv2026

Probing the Mass--Redshift Dependence of Binary Black Holes and its Implications for $H_0$ with GWTC-5.0

The mass and redshift distributions of merging binary black holes (BBHs) bear imprints of their astrophysical formation channels. Whether the BH mass distribution evolves with redshift, however, remains an open question. In this paper, we employ copula functions, which describe the dependence between variables independently of their marginal distributions, to probe the dependence structure between the primary mass and the redshift of BBH mergers. We construct five population models by coupling the marginal distributions of primary mass and redshift with the Gaussian copula, the Clayton copula, or one of its three rotations, and we constrain their hyperparameters using 235 BBH events from GWTC-5.0. For comparison, we also adopt a baseline model in which the primary mass and redshift are assumed independent. Among the copula-based models, only the model with the $180^\circ$ rotated Clayton copula, which couples higher primary masses preferentially with high redshifts, exhibits evidence for a nonzero correlation, with its copula parameter deviating from the independence limit at more than $1σ$ significance. Importantly, the constraints on the Hubble constant ($H_0$) derived from these models are mutually consistent within the $68\%$ confidence level; the baseline model yields $H_0=74.3^{+13.7}_{-19.6}~\mathrm{km\ s^{-1}\ Mpc^{-1}}$, indicating that allowing for a mass--redshift dependence does not significantly bias the $H_0$ inference with the current dataset. Bayesian model comparison favors the baseline model over all copula-based alternatives, with log-Bayes factors corresponding to weak to moderate evidence against the latter on the Jeffreys scale. Our results thus imply that the current GWTC-5.0 sample remains consistent with no redshift evolution of the primary mass distribution over the redshift range probed.

astro-ph.CO↗

arXiv2026

A unified multirate lattice Boltzmann framework for thermosolutal dendritic solidification

Thermosolutal dendritic solidification involves interface evolution, solute diffusion, heat transfer, and melt flow over markedly different time scales. In lattice Boltzmann simulations, a single numerical time interval may place different transport processes in unfavorable relaxation ranges, while asynchronous updates require consistent transfer of phase-change contributions. To address these issues, a unified multirate multiple-relaxation-time lattice Boltzmann method is developed for thermal, solutal, and thermosolutal dendritic solidification. The coupled fields share a common moment-space framework but evolve at different update rates. The concentration and temperature source terms are separated into transport-related and phase-change contributions, and each resolved phase increment is used to immediately transfer the corresponding solutal and latent-heat contributions. The method reproduces characteristic dendritic morphologies and tip-velocity trends under pure diffusion and forced convection, with weak sensitivity to the tested update factors. Directional solidification over Lewis numbers $Le=1$--$1000$ captures the transition from nearly planar to cellular and strongly branched growth, including vertically aligned dendrites and solute-rich interdendritic channels for saline water, in qualitative agreement with experiments. Source-coupling ablation shows that delayed coarse-step transfer produces increasingly strong local source pulses and eventual loss of numerical stability as time-scale separation increases, whereas phase-step instantaneous transfer remains stable over the tested conditions. These results demonstrate the applicability of the proposed framework to dendritic solidification with strongly separated transport time scales. The source code is publicly available in the \emph{DendriteLBM} repository.

physics.flu-dyn↗

arXiv2026

Fysiverse-3D-Vision Technical Report: Generating Executable 3D Worlds from Images through Unified Spatial Reasoning

Generative models have advanced image-conditioned 3D content creation, yet generating controllable and executable 3D scenes from a single image remains challenging. Existing 3D generative approaches can synthesize visually plausible objects and scenes, but their spatial layout estimation is coupled with specific asset generators. They struggle to jointly model object semantics, metric geometry, and scene-level spatial relationships, which are essential for interactive editing, physical simulation, and embodied applications. We propose Fysiverse-3D-Vision, a unified vision-language-geometry framework for generative 3D scene reconstruction and executable asset construction from a single image. We establish a shared representation where spatial reasoning and geometric reconstruction mutually enhance each other, allowing object layouts to be inferred beyond the constraints of individual asset generators. Our model integrates textual supervision, semantic visual cues, and geometric representations within a unified Transformer to capture scene context, metric geometry, and object-level interactions. An object-conditioned layout module performs cross-attention between target object representations and global geometric features to predict object translation, rotation, and scale. Training progressively learns geometry-language alignment, introduces layout reasoning while preserving reconstruction capability, and refines physical consistency through collision-aware optimization. By separating spatial layout reasoning from asset synthesis, Fysiverse-3D-Vision provides an adaptable interface for interactive scene editing, object-level manipulations, and executable 3D content generation. Experiments demonstrate that our framework achieves superior geometric consistency, layout estimation, rendering quality, and physical property understanding compared with existing approaches.

cs.CV↗

arXiv2026

Imperfection for Precision: Upcycling Imperfect Data for High-Precision Robotic Manipulation

Training vision-language-action (VLA) models for high-precision manipulation typically requires task-specific, high-quality data (e.g., teleoperation), which is slow and expensive to collect. To reduce this burden without compromising manipulation precision, we propose $\varepsilon$4P (Imperfection for Precision), a simple yet effective method that "upcycles" two otherwise discarded data sources: (1) low-precision data from the target task and (2) high-precision data from mismatched tasks. Rather than naively mixing these imperfect data sources throughout co-training, $\varepsilon$4P controls where each source contributes along the flow-matching trajectory. Specifically, low-precision, target-task data is used at high noise to preserve high-level task context and high-precision, task-mismatched data is used at low noise to transfer low-level action precision. Through real-robot experiments on both sub-millimeter, high-precision tasks and coarse-grained tasks, we demonstrate that the proposed method (1) effectively leverages additional imperfect data to improve policy performance by up to 31.7 percentage points, and (2) can replace an equal amount of task-specific, high-quality data with an average performance drop of only 4.2 percentage points. Overall, $\varepsilon$4P points toward a scalable paradigm for high-precision manipulation, in which heterogeneous, imperfect data can be systematically repurposed to reduce reliance on costly task-specific, high-quality data. More details are available at https://varepsilon4p.github.io/.

cs.RO↗

arXiv2026

RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention

As the input length of large language model (LLM) serving continues to grow, the KV cache has become a dominant bottleneck in AI infrastructure. It limits GPU memory capacity, serving concurrency, cache reuse, and distributed scalability. Multiple important problems, including position-independent KV cache, prefix KV cache compression, hot/cold KV cache separation, and distributed KV cache management, all depend on how the KV cache is represented and managed. However, existing serving systems largely rely on a monolithic KV cache abstraction, where the KV cache is treated as a homogeneous sequence of token-level memory blocks and managed with similar policies across attention heads and serving scenarios. We observe that KV cache utility is highly structured across KV heads: different heads exhibit different functional roles, attention distances, and runtime importance. Therefore, a full KV cache is not always necessary for every head, token range, or serving scenario. We present RedKnot, a head-aware KV cache management system for LLM serving. RedKnot breaks the conventional monolithic KV cache abstraction by decomposing the KV cache along KV heads, whose importance and effective attention ranges vary significantly across serving scenarios. This head-level decomposition turns the KV cache from a monolithic tensor abstraction into a structured memory object, enabling RedKnot to uniformly support position-independent KV reuse, prefix KV compression, hot/cold KV separation, and distributed KV placement while preserving output fidelity and improving resource efficiency, without requiring model retraining or fine-tuning. RedKnot establishes a new foundation for AI infrastructure by transforming the KV cache from a monolithic, passive runtime artifact into a dynamic, model-aware runtime substrate for scalable LLM serving.

cs.AI↗

arXiv2026

Bayesian Optimization of The Relativistic Heavy Ion Collider Luminosity via $s^*$ Control

Maximizing luminosity at the interaction point (IP) requires the collision location $s_{IP}$ to coincide with the longitudinal position of the minimum beta function, $s^$. Accurate optics measurements and control of $s^$ are therefore essential for luminosity optimization. At the Relativistic Heavy Ion Collider (RHIC), average horizontal beta-beat measurements between operating IPs are approximately $20%$, with significant variation in measured $s^*$. Precise control of the beam waist position $s^$ is particularly challenging for modern high-energy colliders with short bunch lengths and large crossing angles. We present an online Bayesian optimization (BO) application using the GPTune framework to optimize sPHENIX luminosity through $s^$ control at RHIC. GPTune was first validated at the RHIC Electron Beam Ion Source (EBIS), where it achieved up to a $70%$ increase in beam intensity over the baseline, although experienced operators could reach similar performance with longer manual tuning. The framework was subsequently deployed during sPHENIX operations. Using an intensity-normalized Zero-Degree Calorimeter (ZDC) signal as the optimization objective, due to the unavailability of the live sPHENIX MVTX signal, GPTune identified local luminosity maxima, recovered from intentionally degraded $s^$ configurations, and revealed residual horizontal and vertical waist offsets in the interaction region. These results demonstrate the robustness and efficiency of Bayesian optimization for real-time collider tuning under noisy, time-varying conditions. The $s^$ control methodology provides a promising tool for precision luminosity optimization and is particularly relevant to next-generation short-bunch colliders such as the Electron-Ion Collider (EIC).

physics.acc-ph↗

arXiv2026

Fysiverse-3D-SimReady Technical Report: Agentic Physical Simulation for Pragmatic 3D World Reconstruction

Agentic recognition requires visual perception to move beyond static scene understanding and produce structured scene representations that support the perception--reasoning--action loop. Existing single-image 3D generation methods, however, mainly produce visually plausible object assets rather than simulation-ready scene states. When independently generated meshes are composed in a shared space, they may fail to align with the input camera, violate gravity, interpenetrate nearby objects, or become unstable under physics simulation. We present Fysiverse-3D-SimReady, a grounded refinement framework for reconstructing simulation-ready multi-object scenes from a single RGB image with instance and ground prompts. The method places generated object meshes into a shared gravity-aligned scene, refines their camera-space poses through differentiable rendering, and corrects scene-level supports and contacts for stable physical execution. The scene is then used by an agentic simulation workflow, which converts a scene-specific task goal into an executable physics rollout rendered from the original camera view. Experiments show that Fysiverse-3D-SimReady improves input-view alignment, contact plausibility, and physical stability over existing single-image reconstruction and scene generation baselines, while enabling goal-conditioned physical interactions from a single image.

cs.CV↗

arXiv2026

LG-PF: Lightweight Confidence-Guided Polarization Image Fusion

Polarization image fusion combines the stable luminance and structural information of the total- intensity image S0 with the material-sensitive details of the degree of linear polarization (DoLP) image. However, the reliability of DoLP varies spatially, and indiscriminate polarization transfer may amplify unstable responses or disturb the structural appearance anchored by S0. We therefore propose LG-PF, a lightweight confidence-guided framework that formulates polarization fusion as a selective residual transfer process. A Polarization Confidence Prior estimates spatially reliable polarization responses, a Mask-guided Multi-scale Fusion module regulates their transfer across three feature scales, and a Lightweight Context-aware Bounded Correction Head stabilizes local photometric and structural transitions. Confidence guidance is also incorporated into the optimization objectives to preserve reliable polarization details while suppressing unsupported responses. We also construct MSP, a multi-scene polarization fusion dataset containing 1000 pixel-aligned image pairs from 17 indoor and outdoor scene categories. LG-PF achieves the best results across all six evaluated metrics on MSP, while subset-based evaluations on PIF and GAND show promising transferability without fine-tuning. With only 0.2936 M parameters and an inference time of 21.712 ms per image, LG-PF achieves competitive fusion quality with low computational cost. The source code will be available at https://github.com/1hzf/LG-PF.

cs.CV↗

arXiv2026

LGFN: Lightweight Gated RGB-Polarization Fusion with Modality-Availability Conditioning for Camouflaged Object Detection

Camouflaged object detection (COD) is an important engineering task in intelligent optical perception, but it remains challenging when targets closely resemble their surroundings. Polarization imaging provides complementary physical cues, whereas existing methods typically assume fixed multimodal input configurations and entangle intra-polarization coordination with interaction between red-green-blue (RGB) and polarization representations. We propose LGFN, a lightweight gated RGB-polarization fusion framework supporting separately optimized RGB-only and polarization-assisted configurations. A deterministic Modality Router selects the appropriate configuration according to polarization availability. In the multimodal configuration, an availability-conditioned Modality Gate calibrates the available polarization branches; the Gated Polarization Hub coordinates learned degree of linear polarization (DoLP) and angle of polarization (AoP) representations with explicit polarization cues; and RGB-Polarization Cross Fusion introduces the coordinated representation into the RGB hierarchy through controlled residual interaction. The multimodal configuration requires neither sample-dependent statistics nor handcrafted quality descriptors during inference. On the complete 230-image PCOD_1200 test set, the RGB-only configuration achieves a mean absolute error of 0.0090, a Dice score of 0.8806, and an intersection over union of 0.8144, obtaining the best results on all six metrics among the evaluated RGB-based methods. Under a common local reevaluation protocol, the multimodal configuration outperforms PolarNet and IPNet on all six metrics. Relative to IPNet, it reduces the parameter count, floating-point operations, and latency by 53.1%, 73.6%, and 63.0%, respectively. The source code will be available at https://github.com/1hzf/LGFN.

cs.CV↗

arXiv2026

OmniRisk: Omnidirectional Trajectory-Risk Learning for Agile Quadrotor Dynamic Avoidance

Agile quadrotor avoidance of fast-moving obstacles requires anticipating collisions and selecting feasible maneuvers within short reaction windows. Reliable predictive avoidance remains challenging because sparse range observations do not directly reveal obstacle motion, while online trajectory optimizers either scale poorly with obstacle count or remain efficient at the expense of reliability in dense, high-speed encounters. We present OmniRisk, an omnidirectional planning framework that learns trajectory-level risk offline for efficient onboard evasion. A fixed-dimensional tensor combines LiDAR range panoramas, dynamic masks, and Cartesian surface velocities to represent geometry and motion jointly. We formulate an asymmetric risk field aligned with obstacle velocity that emphasizes approaching interactions and attenuates receding ones. Accumulating this risk along predicted relative trajectories provides dense supervision and discourages unnecessary hesitation after obstacles pass. A dual-branch circular convolutional network predicts terminal boundary states and dynamic risks for candidate primitives over an omnidirectional anchor lattice in a single forward pass, followed by selection and closed-form reconstruction of the selected candidate primitive. This formulation removes online risk accumulation along trajectories and makes risk-inference cost independent of obstacle count. OmniRisk enables efficient onboard avoidance, with real-world flights demonstrating consecutive evasive maneuvers at relative encounter speeds up to 15 m/s without fine-tuning. Code is available at https://github.com/VANdexj/OmniRisk.

cs.RO↗

arXiv2026

Can Multimodal Large Language Models Generate and Detect Multimodal Social Media Fake News?

The rapid advancement of generative AI raises concerns about the misuse of Multimodal LLMs (MLLMs) for large-scale disinformation campaigns on social media. Despite existing research on textual disinformation, a fundamental question remains unanswered: can MLLMs be exploited to fabricate realistic multimodal fake news, and can they reliably detect it? We introduce a multi-agent framework in which a story agent, an image agent, and a critic agent collaborate to produce fake social media posts that plausibly counter true news. We apply the framework to generate over 9,000 paired multimodal news posts across science, health, and entertainment domains, and benchmark 16 open- and closed-source MLLMs for automated detection. We find that most models fall substantially short of human-level accuracy and fail critically on identifying image authenticity. Our research provides a foundation for developing robust defenses against social media fake news. Code and data are available at https: //github.com/xiuzhenzhang/Multimodal.

cs.CL↗

arXiv2026

On Kernel Eigen-alignments of KRR: Reconstruction and Generalization

This paper investigates the critical role of eigenalignments between the kernel matrix and learning targets in achieving robust generalization in learning problems. We establish a direct connection between generalization performance in kernel methods and the estimation of eigenvectors and eigenvalues of matrices, offering a more intuitive understanding compared to prior work with minimal assumptions. We also show that, since the prediction task in KRR is essentially the weighted sum of eigenvectors/singular vectors, by analyzing how much error can be caused by perturbations to the kernel matrix, we can then derive a bound on this generalization error using the estimation stability of matrix eigenvalues and eigenvectors. Compared with previous work, our analysis concentrates on finite-sample settings and on the generalization error arising from having a suboptimal finite training set. Our findings reveal that in kernel methods, as long as the kernel is of high rank, the near-zero reconstruction error can be trivially obtained, implying that the reconstruction error will have limited predictive power for generalization. Additionally, we establish a generalization bound from an eigenvalues/eigenvectors estimation perspective, showing that strong generalization requires increasing eigenvector alignment, eigenvalue magnitude, or gaps between consecutive eigenvalues. Finally, we also demonstrate empirically our simplified risk estimator is able to track testing errors in trailing eigendirections more closely than the existing literature.

stat.ML↗