arXiv subjects

Yang Liu

Publications and source records attributed to Yang Liu.

At least 199 records · Page 11Linked to original sources

arXiv2026

ReMemNav: Memory-Based Decision Correction and Target Verification for Zero-Shot Object Navigation

Zero-shot object navigation requires agents to locate unseen targets in unfamiliar environments without prior maps or task-specific training. Despite the commonsense reasoning ability of vision-language models (VLMs), existing mapless navigators often suffer from repeated exploration and premature stopping caused by limited historical context and false-positive target predictions. We propose ReMemNav, a training-free framework that combines lightweight semantic grounding, memory-based decision correction, and target verification. RAM-derived semantic priors guide panoramic direction selection, while a geometry-triggered correction mechanism retrieves historical scene descriptions when the selected direction points toward previously visited regions. When a target is predicted to be visible, ReMemNav verifies the corresponding single-view observation before the final approach. Depth-based action sampling then converts the high-level decision into collision-free motion. Experiments on HM3D and MP3D show that ReMemNav achieves higher success rates and path efficiency than existing training-free zero-shot baselines. Specifically, ReMemNav achieves absolute SR/SPL gains of 1.7%/5.3% on HM3D v0.1, 18.2%/11.1% on HM3D v0.2, and 8.7%/6.9% on MP3D.

cs.RO↗

arXiv2026

On passive recovery of structured elastic density and initial states

We study simultaneous recovery of the (variable) mass density, initial displacement, and initial velocity for the three-dimensional isotropic elastic wave equation with known constant Lamé parameters. The data are the complete displacement trace on an enclosing boundary. We first assume that the density-weighted initial displacement and velocity have fixed known profiles in one spatial direction. The $s^0$ and $s^1$ coefficients of the zero-frequency Laplace expansion identify these weighted states. The $s^2$ and $s^3$ coefficients then give static Lamé orthogonality identities for the density difference. The exact difference expansion through order $s^3$ has an $O(s^4)$ remainder, uniformly on bounded spatial sets and bounded density-contrast and weighted-state classes. Under alignment of the two initial states and a nonzero moment of the density-weighted initial velocity, the two density identities reduce to a constant-vector static transform. Two opposite elastic null phases give uniqueness for one or two fixed vertical density profiles when the profile family is two-sided Laplace nondegenerate. When $λ+μ\ne0$, each nonzero normal root has partial multiplicities $2$ and $1$ and admits a length-two Jordan chain. The associated polynomial--exponential Lamé mode produces derivatives of the bilateral profile transforms. The resulting Hermite--Laplace system gives uniqueness for aligned classes with up to four fixed vertical profiles and independent horizontal coefficients, provided that the profile system is Hermite--Laplace nondegenerate. Distinct translations of one compactly supported profile provide an explicit four-profile class. We also prove rigidity of the alignment reduction and exhibit an infinite-dimensional kernel for the reduced transform on unrestricted densities.

math.AP↗

arXiv2026

SaFeR-Steer: Evolving Multi-Turn MLLMs via Synthetic Bootstrapping and Feedback Dynamics

MLLMs are increasingly deployed in multi-turn settings, where attackers can escalate unsafe intent through the evolving visual-text history and exploit long-context safety decay. Yet safety alignment is still dominated by single-turn data and fixed-template dialogues, leaving a mismatch between training and deployment. To bridge this gap, we propose SaFeR-Steer, a progressive multi-turn alignment framework that combines staged synthetic bootstrapping with tutor-in-the-loop GRPO to train a single student under adaptive, on-policy attacks. We also introduce Trajectory-Consistent Summative Reward (TCSR), which aggregates the historical minimum and average of turn rewards so that any low-quality turn affects the trajectory-level return. I. Dataset. We release STEER, a multi-turn multimodal safety dataset with STEER-SFT (12,934), STEER-RL (2,000), and STEER-Bench (3,227) dialogues spanning 1-10 turns. II. Experiment. Starting from Qwen2.5-VL-3B/7B, SaFeR-Steer substantially improves Safety/Helpfulness on both single-turn (48.30/45.86 $\rightarrow$ 81.84/70.77 for 3B; 56.21/60.32 $\rightarrow$ 87.89/77.40 for 7B) and multi-turn benchmarks (12.55/27.13 $\rightarrow$ 55.58/70.27 for 3B; 24.66/46.48 $\rightarrow$ 64.89/72.35 for 7B), shifting failures to later turns and yielding robustness beyond scaling alone. Code is available at https://github.com/Ed-Bg/SaFeR-Steer-full

cs.LG↗

arXiv2026

OphBiWSSD: Scaling Temporal Action Localization in Ophthalmic Surgeries with Bidirectional Weight-tied State Space Duality

High-frequency surgical maneuvers in ophthalmology necessitate high-fidelity temporal modeling, yet characterizing long-range procedural dependencies remains computationally prohibitive for attention-based architectures. Existing models often require aggressive temporal downsampling, which compromises the detection of fine-grained action boundaries and instrument-tissue interactions. To address these scalability constraints, we present OphBiWSSD, a framework that reformulates surgical temporal action localization leveraging Bidirectional State Space Duality. By employing a weight-tied selective scan mechanism that incorporates both preceding and succeeding surgical contexts, our approach facilitates the global synthesis of non-causal temporal cues with linear complexity. This streamlined architecture is well-suited to capture the bidirectional dependencies present in ophthalmic workflows, effectively bridging the gap between local boundary precision and long-range procedural context without incurring the quadratic memory overhead of traditional Transformers. Extensive experiments on the OphNet benchmark demonstrate that OphBiWSSD achieves state-of-the-art temporal localization performance, with mean Average Precisions of 44.42% on phases and 43.08% on operations, surpassing the baselines by 6.80% and 6.66%, respectively. Empirical validation indicates that our approach ensures precise temporal localization and offers a computationally viable pathway for deploying surgical intelligence systems in clinical environments. The code is publicly available at https://github.com/yo3nglau/OphBiWSSD.

cs.CV↗

arXiv2026

An Ultra-Widefield Swept-Source OCTA Dataset and a Polar-Gated Mamba Network for Retinal Vessel Segmentation

Ultra-widefield (UWF) swept-source optical coherence tomography angiography (SS-OCTA) enables large-area retinal vascular imaging, yet vessel segmentation at this scale lacks dedicated public benchmarks and comprehensive evaluation for quantitative vascular analysis. We introduce WOIVES, to our knowledge the first publicly available UWF SS-OCTA vessel-segmentation dataset, comprising 206 eyes from 152 participants with a 24x20mm^2 field of view. WOIVES spans emmetropia to high myopia and provides soft probability vessel annotations. We further propose PG-Mamba, a visual state space model that enhances conventional directional scans with two complementary polar-coordinate scan orders. An auxiliary Dynamic FOV Gating module performs spatial modulation at the bottleneck. PG-Mamba outperformed seven competitive approaches on broad segmentation metrics under cross-validation. It achieved the lowest median absolute errors for vessel density, fractal dimension, and vessel length density. WOIVES is publicly available on Zenodo (DOI: 10.5281/zenodo.21904672), and the PG-Mamba code is available at https://github.com/syb1234567/PG-Mamba.

cs.CV↗

arXiv2026

Impact waves in soft bilayer tissues

In this study, we investigate impact-wave propagation in a pre-stressed compressible hyperelastic bilayer resting on a frictionless rigid substrate within the framework of nonlinear elasticity. Semi-analytical solutions for the transient displacement fields induced by localized surface impulses are derived using the Fourier-Laplace transform, and their long-time asymptotic behavior is obtained through the method of stationary phase. Finite element simulations are further performed to validate the theoretical predictions. The results establish a direct connection between characteristic features of the transient displacement fields and the group velocities of the guided-wave modes, revealing how dispersive propagation and modal contributions shape the observed waveforms. Three representative bilayer models corresponding to human articular cartilage, gingiva, and skin are then considered to examine the effects of variations in layer stiffness and thickness. The results show that structural and material variations can substantially modify the spatial distribution, arrival positions, and characteristic velocities of impact waves. In particular, higher-order modes can make significant contributions to the displacement response in certain parameter regimes, such that the dominant features of the wavefield cannot always be characterized by the fundamental mode alone. These findings demonstrate that transient-wave measurements contain mechanical information beyond that captured by a single characteristic surface-wave velocity and provide a theoretical basis for interpreting transient wavefields in multilayered soft materials and for future non-invasive mechanical characterization.

cond-mat.soft↗

arXiv2026

Personalized Execution Time Optimization for Billion-Scale Scheduled Jobs

Scheduled batch jobs are widely used on asynchronous computing platforms to execute enterprise applications such as promotional notifications and candidate pre-computation for recommender systems. Delivering or updating information at the right time is important for user experience and execution impact, yet providing a versatile, personalized execution time optimization solution across diverse product scenarios while maintaining reasonable infrastructure costs remains challenging. In this paper, we present a deployed system that serves billions of users daily, combining learning-to-rank with a "best time policy" for execution time selection. We describe the four-stage evolution of our approach: from heuristic peak-hour rules, to pointwise ML-based activity pattern predictions, to a linear signal assembler with globally fixed weights, and finally to a contextual ensemble learner that produces per-user adaptive fusion weights via a neural policy network trained with listwise learning-to-rank objectives. We further report the discovery of cross-use-case cannibalization effects and introduce a coordination system to mitigate the problem. Our production experiments demonstrate measurable improvements in both execution efficiency and downstream product impact. We share deployment lessons including failure analyses and design decisions accumulated over four years of operating this system at scale. To our knowledge, this represents the first ML-based multi-tenant execution time optimization system deployed across different product domains at industrial scale.

cs.LG↗

arXiv2026

Benchmarking Requirement-to-Architecture Generation with Hybrid Evaluation

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

cs.SE↗

arXiv2026

TEFM: Token-Efficient Faithful Modeling for Structured Data

In this paper, we solve two fundamental obstacles in applying LLMs to critical domains: token efficiency and faithfulness. To address both constraints jointly, we present TEFM (Token-Efficient Faithful Modeling), a framework designed for structured data analysis in critical domains. TEFM achieves token efficiency by compressing lengthy structured observations into compact Behavioral Code tokens, dramatically reducing token consumption with minimal information loss. Moreover, TEFM enables faithful rationalization through a dual-fidelity objective that jointly optimizes code-level reconstruction and prediction-level fidelity, identifying minimal sufficient feature subsets grounded in input data. Comprehensive experiments across various domain datasets and model backbones (Qwen3, Gemma-2, Phi-4) show that TEFM achieves competitive classification accuracy with dramatic token reduction (approximately 1\% token retention in clinical and 2\% in security domains) while producing faithful rationales.

cs.CL↗

arXiv2026

A Trust-Network-Based Federated Learning Framework for Multi-Center Aging Clock Prediction

Aging clocks quantify biological aging and help characterize individual health status. What protein interactions are important for accurate aging clocks, and are they zeroth-order or higher-order? Addressing these questions requires learning from large molecular datasets distributed across medical centers, where privacy constraints prevent centralized data sharing. Federated learning offers a natural solution but faces four challenges in this setting: limited local sample sizes, sparse and directional inter-center trust, the need to retain discriminative age prediction while supporting interpretation, and model drift and forgetting under heterogeneous cross-center data. We propose TNFL, a trust-network-based federated learning framework that progressively propagates models along directed pairwise trust relations without centralized aggregation. TNFL combines an age-aware mixture-of-experts model with generative replay to preserve previously learned information and reduce forgetting and drift. Experiments across multiple molecular datasets show that TNFL enables effective aging-clock prediction with limited local data, provides interpretable age-dependent prediction patterns, and maintains stable performance across interaction orders. To investigate the biological questions, we analyze TNFL-identified pairwise protein interactions and their higher-order organization through functional and network analyses. The identified interactions repeatedly form coordinated higher-order subnetworks spanning multiple aging-related biological systems, with several proteins recurring across subnetworks. These findings suggest that TNFL captures molecular relationships beyond isolated pairwise associations and reveals coherent higher-order biological organization associated with aging.

cs.LG↗

arXiv2026

Tightening Bounds on Warm Dark Matter with High-Redshift Gamma-Ray Bursts

The cold dark matter paradigm successfully explains large-scale structure but faces persistent tensions on small scales. Warm dark matter (WDM) with $\mathrm{keV}$-scale particles can alleviate these issues by suppressing small-scale structure formation. The presence of collapsed structures at high redshifts places strong lower limits on the WDM particle mass $m_x$. Gamma-ray bursts (GRBs) are ideal high-redshift probes due to their extreme brightness. Using the most recent \emph{Swift} GRB data accumulated over the past two decades, we derive robust constraints on $m_x$ by conservatively assuming that the comoving GRB formation rate is proportional to the cosmic star formation rate (SFR), with an additional redshift evolution parameterized as $(1+z)^α$. Applying a maximum-likelihood analysis to 118 GRBs with redshift $z<10$ and luminosity $L\ge 4.0\times10^{52}\,\mathrm{erg\,s^{-1}}$, we obtain $m_x \gtrsim 1.23\,\mathrm{keV}$ and $α=1.57^{+1.14}_{-0.57}$ at the 95\% confidence level (CL). The no-evolution scenario ($α=0$), in which the GRB rate exactly traces the SFR without additional redshift evolution, is excluded at the $5σ$ level. Adopting the best-fit value $α=1.57$ as a prior tightens the lower limit on $m_x$ to $m_x \gtrsim 1.59\,\mathrm{keV}$ at the 95\% CL. These robust constraints demonstrate that GRBs are a powerful probe of the early Universe. A better understanding of the relationship between the GRB rate and the SFR would enable even tighter limits on WDM models.

astro-ph.HE↗

arXiv2026

Neural Centroidal Voronoi Tessellations

Centroidal Voronoi tessellation (CVT) is a fundamental primitive for high-quality surface sampling and isotropic remeshing in computer graphics. However, computing surface CVTs with classical solvers remains expensive: each optimization step repeatedly constructs restricted Voronoi diagrams (RVDs) and integrates quantities over their surface cells. We introduce Neural CVT, a learning-based surface-CVT solver that replaces these costly geometric computations with a recurrent neural optimizer, accelerating CVT optimization by one to two orders of magnitude in our benchmarks while preserving geometric fidelity. Given an input surface, we sample a dense point cloud and extract multi-scale geometric descriptors with a graph neural encoder. A lightweight recurrent optimizer then refines seed positions over a small number of iterations, aggregating interpolated surface features and optimization history to predict per-seed displacements. The framework is trained self-supervised using CVT objectives that promote uniform distributions and, when desired, feature alignment. Across diverse organic and CAD-like shapes, Neural CVT generalizes to unseen geometry, initialization strategies, and seed densities, producing isotropic, feature-preserving remeshes comparable to state-of-the-art offline optimization methods at a fraction of the computational cost. Code and trained models will be released.

cs.GR↗

arXiv2026

AuK Technical Report: An Open-Source Foundational Model for Speech Generation and Editing

We introduce AuK, an open-source foundational model that unifies speech generation and editing through a common interface of natural-language instructions and audio context. To support this broad capability set, we construct approximately 3.03 billion instruction--audio instances and 1.95 million hours of effective supervision across five task families: speech generation, content editing, enhancement and separation, paralinguistic editing, and acoustic editing. AuK combines a multimodal large language model for semantic conditioning, an VAE jointly trained on speech, general audio, and music for acoustic conditioning, and a hybrid rectified-flow Transformer that performs dual-stream MMDiT blocks followed by unified single-stream DiT blocks for generation. Training begins with generation-only warm-up and proceeds to joint generation--editing pre-training. We then apply complementary post-training strategies: human-feedback preference optimization for open-ended editing and reward-based reinforcement learning for speech generation. To reduce inference cost, we further distill the model with consistency initialization and task-routed Decoupled DMD. The resulting AuK-Flash performs 4-step inference without classifier-free guidance and achieves a 4.5 wall-clock speedup over the full model under matched conditions. Experiments demonstrate leading performance on zero-shot and instruction-controlled speech generation and general instruction-guided editing, while remaining competitive on signal-level restoration tasks. We release both the source code and model weights to support reproducibility and further research.

cs.SD↗

arXiv2026

Studying Image Tokenizers as Visual Languages in Unified Multimodal Models

Image tokenizers define the ``visual language'' of unified multimodal models, yet are commonly studied through isolated metrics or generation-/understanding-only evaluations. These evaluations do not fully capture how visual tokens behave when modeled jointly with text. We build a controlled pure-autoregressive testbed and track task-specific validation losses during multimodal continual pretraining across text, image, text-to-image (T2I), and image-to-text (I2T) prediction. We examine how these losses scale and relate to downstream performance, then use them to study multimodal learnability---how well image and text tokens are jointly modeled---and tokenizer design. We find that (1) losses should be analyzed by task, since they exhibit distinct scaling behavior and rank tokenizers differently. (2) The loss--performance relationship depends on the predicted token space: for a fixed tokenizer, T2I and I2T losses correlate with generation quality, but across tokenizers, the T2I loss--performance relationship shifts with the image-token space, whereas I2T loss, computed over a shared text vocabulary, provides a more consistent signal. I2T loss also correlates with both generation and visual understanding performance after supervised finetuning. Using losses as a lens, we show that (3) better reconstruction does not necessarily yield lower task-specific losses or stronger downstream performance, and that (4) image tokenizer choice can affect text modeling under joint optimization. As case studies, we revisit three tokenizer design axes---the discriminator, semantic supervision, and vocabulary size---to examine their effects on joint modeling and downstream performance. Together, our testbed offers a complementary perspective on image tokenizers as visual languages, highlighting their interplay with text in joint multimodal training.

cs.CV↗

arXiv2026

Low-Rank Dynamics-Effective Latent Carriers for Counterfactual Rollout in Learned World Models

We ask whether a small, directly addressable hidden-state intervention can place a learned world model on an intended counterfactual future and then let the model's own dynamics carry that future forward. In a controlled two-object collision environment, we study a 192-dimensional recurrent model trained on factual and locally edited counterfactual trajectories. Candidate carriers are learned from training-only counterfactual-minus-factual hidden differences, and an affine map predicts carrier coordinates from the factual state and requested edit without access to the native counterfactual hidden state at test time. For bounded single-component velocity edits, rank 4 is the smallest tested rank on the preregistered grid that satisfies the development criteria. A one-shot rank-4 patch launches a 12-transition autonomous rollout without future observations, teacher forcing, repeated hidden-state correction, or physical-state clamping. The frozen procedure satisfies the preregistered 2-of-3 fresh-checkpoint replication rule and remains reusable at nearby anchors. The same Single-derived carrier and Single-only affine map also support bounded same-object two-component requests. Across the matched training regimes, broader counterfactual support was associated mainly with better Joint rollout accuracy and more additive Joint hidden responses. Composition-related structure is enriched in the rank-4 subspace but is not confined to it, and local recurrent diagnostics show strong one-step coupling from the carrier to the rest of the hidden state. A position-edit stress test fails the required specificity controls. Together, these results support a compact dynamics-effective intervention-entry interface, not a closed four-dimensional state or an intrinsic state dimension.

cs.RO↗

arXiv2026

WM-Craftnet: World Synesthesia Model for Generalizable and Robust Dexterous In-Hand Manipulation

Generalizable and robust dexterous in-hand manipulation requires a policy to infer object pose, geometry, contact, and potential slip from partial and noisy observations. Although recent tactile and visuotactile RL methods achieve strong in-hand rotation in controlled settings, their robustness often degrades under pose shifts, force disturbances, and object variation. We propose WM-Craftnet, a world-model-conditioned framework that learns compact action-conditioned latent dynamics from proprioception, depth, tactile sensing, and actions, supervised by multimodal reconstruction and reward prediction. Rather than using the world model for latent imagination or policy optimization, WM-Craftnet uses the learned World Synesthesia Model (WSM) as recurrent task context for an asymmetric actor--critic policy. Importantly, WSM is trained to reconstruct clean depth targets from noisy depth inputs, providing a denoised geometric state for real-robot deployment. Ablations over recurrent baselines, auxiliary heads, tactile masking, and WSM modality heads show that predictive world modeling, clean-depth supervision, and tactile contact cues all shape the learned state. A WSM pretrained on nine $z$-axis objects serves as a reusable prior for $49$-object downstream policy learning. This context improves multi-object rotation, with quantitative and qualitative evidence for unseen-object, perturbation-recovery, and sim-to-real transfer.

cs.RO↗

arXiv2026

Elastic Horizon: Discovering the Effective Interaction Frontier in Agentic Reinforcement Learning

Scaling the interaction horizon-the maximum number of environment interactions per episode-improves LLM agents on long-horizon tasks, and curriculum-based methods that progressively expand the horizon outperform fixed-horizon alternatives. However, existing schedules are open-loop: they monotonically increase the horizon until a manually specified maximum, with no mechanism to detect when further expansion stops helping. We propose the effective interaction frontier hypothesis: a dynamic boundary beyond which additional interactions yield diminishing returns while cost grows linearly. We then introduce Elastic Horizon, a closed-loop controller that tracks this boundary via the 90th percentile of successful trajectory lengths. On AppWorld and BFCL, fixed-horizon sweeps reveal clear saturation plateaus; Elastic Horizon stabilizes the horizon inside the saturation band from both under- and over-capacity initializations, attains the best success rates across 7B and 14B backbones, and saves up to 25% of per-step trajectory tokens. Our work shifts the paradigm from how to scale interaction horizons to when to stop scaling.

cs.AI↗

arXiv2026

Trajectory Optimization via Schrödinger Bridge Sampling

We take a new look at the relation between finite-horizon trajectory optimization and Schrödinger bridge sampling. Viewed as inference, KL-regularized trajectory optimization is solved by sampling from a Gibbs--Boltzmann distribution whose energy is the trajectory cost, and the adjoint Schrödinger bridge sampler (ASBS) is a simulation-free diffusion sampler designed for exactly such unnormalized targets. Hard equality path and terminal constraints, by contrast, confine the admissible decision variables to a measure-zero feasibility manifold, on which the target must be redefined intrinsically. In particular: $(a)$ we analyze two complementary parametrizations; a rollout parametrization, in which the dynamics are eliminated and only the remaining constraints shape the manifold, and a double-shooting parametrization, in which states and controls are sampled jointly and the dynamics themselves become part of the manifold; $(b)$ we establish regularity conditions under which both admissible sets are smooth embedded manifolds; $(c)$ under compactness and path-connectedness assumptions, we sample from the resulting intrinsic Gibbs measures via Riemannian ASBS, treating strict inequalities through exponential slack variables. Experiments, including contact-rich locomotion and manipulation, demonstrate the effectiveness of both regimes.

eess.SY↗