Searcharxiv⌕ Search

arXiv subjects

Ye Tian

Publications and source records attributed to Ye Tian.

At least 19 recordsLinked to original sources
arXiv2026

Harnessing Intrinsic Subject-Aware Attention for Controllable Multi-Subject Video Generation

Multi-subject video generation faces two key challenges: uncontrollable fidelity strength and potential semantic drift. We address these by analyzing the internal mechanisms of Diffusion Transformers (DiTs). We found that certain attention blocks naturally form an Intrinsic Spatial Grounding Map (ISGM) that precisely locates reference subjects. Building on this insight, we propose Dual-phase Intrinsic Attention Leveraging (DIAL), a framework that uses these internal signals for both training and inference. In low-noise stages, we use ISGM to guide the attention mechanism, allowing precise control over fidelity strength during inference without retraining. In high-noise stages, we use these same maps to automatically build preference pairs at no additional cost for Reinforcement Learning (RL). This RL procedure effectively anchors the model's attention to reference subjects and mitigates semantic drift. Extensive experiments show that DIAL significantly outperforms baseline models on the OpenS2V-Eval benchmark, consistently improving identity consistency and enabling controllable fidelity strength.

cs.CV↗
arXiv2026

FLARE: A Failure-Aware Framework for Autonomous Correction and Recovery in Visual-Language Robotic Manipulation

Vision-Language-Action Models~(VLAs) have demonstrated significant promise in generalizing to complex, long-horizon robotic manipulation tasks. However, their performance remains brittle, as they are typically trained on trajectory-monotonic, failure-free demonstrations. This reliance on ``perfect" data leaves them unable to recover from common execution errors, such as a missed grasp, a dropped object, or an unexpected collision. In this paper, we propose FLARE, a novel framework that endows VLAs with robust error recovery capabilities through a ``Retry" and ``Reset" paradigm. First, we introduce a ``Retry" mechanism by injecting perturbation and bridging segments that decouple robot pose from environment state into demonstrations, enabling the policy to autonomously handle execution deviations. Second, to address critical, state-breaking (OOD) failures, we introduce a ``Reset" pipeline. We leverage an MLLM for offline failure analysis to automatically identify OOD states from execution videos. This analysis enables the efficient, targeted collection of a small library of object-centric ``Reset" skills, which are trained to restore the environment to a task-valid state. Our full framework integrates these learned policies. At inference, an online MLLM monitor arbitrates between task execution and ``Reset" skills. Experiments on challenging, contact-rich manipulation tasks show our approach significantly improves task success and robustness.

cs.RO↗
arXiv2026

On-Policy Self-Distillation in Diffusion Models

Reinforcement learning can align diffusion models with human preferences and task-specific objectives, but endpoint rewards do not specify how an intermediate denoising prediction should change. We introduce DiffusionOPSD as an on-policy self-distillation framework that converts image-level reward guidance into explicit targets for clean-output predictions at sampled queries. At each outer iteration, a frozen behavior policy generates trajectories and supplies query states and anchors. Reward gradients construct bounded positive and negative targets around each anchor. The trainable policy fits these targets as detached supervision through finite fitting before an exponential moving average update refreshes the behavior policy. This setup lets us measure target construction and finite realization separately. Controlled same-query experiments show that larger target-construction gains do not necessarily translate into larger realized gains after a single fitting update. Across SD 3.5-M and the step-distilled Z-Image-Turbo, our approach achieves the best final held-out scores in 19 of 20 reward-matched settings across two backbones and ten evaluators. It outperforms the strongest competing method by up to 44.0% and reduces training GPU-hours relative to DiffusionNFT by 40% on SD 3.5-M and 63% on Z-Image-Turbo. These results support on-policy self-distillation as an efficient and analyzable approach to diffusion post-training by converting image-level reward guidance into explicit and continually refreshed intermediate supervision, thereby opening a path toward more efficient and diagnosable alignment.

cs.CV↗
arXiv2026

HiPHI: A Large-Scale Benchmark for High-Precision Human Motion and Object-Interaction

Humanoid intelligence requires learning over an extremely diverse space of whole-body motions and physically grounded interactions. However, existing embodied datasets remain fundamentally limited: internet-scale video data lack precise physical states and interaction grounding, while laboratory motion datasets provide high fidelity but only narrow behavioral coverage. This mismatch creates a critical bottleneck for scalable humanoid policy learning. We present HiPHI, a 600+ hour scale high-fidelity whole-body human motion dataset designed to systematically maximize coverage of the human motion and interaction manifold. HiPHI is theoretically guided by FrameNet, a linguistic framework organizing human primitives. Created using an optical motion capture pipeline, HiPHI provides sub-millimeter spatial marker tracking accuracy for full-body human motion and mesh-level object trajectories. We further introduce a benchmark suite evaluating motion-space diversity, interaction grounding, object consistency, and physical AI applications. Our analyses demonstrate that HiPHI significantly expands motion coverage compared to existing motion datasets while maintaining high-fidelity interaction quality, and establishes a scalable data foundation for training, evaluating, and generalizing humanoid policies in real-world embodied tasks, where similar extensions are also applicable to motion prior models in computer graphics. Project page: https://noitom-robotics.github.io/hiphi/

cs.RO↗
arXiv2026

Sufficient Dimesion Reduction via Generalized Stein's Lemma

Sufficient dimension reduction (SDR) seeks the minimal subspace of the predictors that captures the full conditional distribution of the response, which is known as the central subspace (CS). When the response is multivariate, the problem becomes considerably more challenging, particularly when the sample size is limited. Existing methods face different limitations:inverse regression approaches rely on strong distributional assumptions and matrix inversion, and their multi-response extensions suffer from severe slice sparsity; forward regression methods depend on computationally intensive iterative smoothing whose cost grows with the response dimension; and deep learning-based approaches demand large amounts of labeled data. To circumvent these shortcomings, we propose an SDR framework based on the generalized Stein's lemma. Our method constructs a cross-moment matrix between the multivariate response and the marginal score function of the predictors, and recovers the CS via its singular value decomposition. The proposed method does not rely on the linearity condition, avoids matrix inversion as well as iterative smoothing, and can leverage unlabeled data when available. We establish convergence guarantees for the proposed estimator under standard regularity conditions. Moreover, we propose a practical rank-selection algorithm to estimate the dimension of the CS. Extensive simulation studies and a real data application demonstrate that the proposed methods consistently outperform existing approaches across a variety of settings, particularly in moderate-dimensional, label-scarce scenarios with high noise levels.

stat.ML↗
arXiv2026

High Precision Fundamental Physics Experiments at JLab with Spin-transparent Storage Rings of Low-energy Polarized Electron Beams

A breakthrough in fundamental physics experiments measuring particle spin precession may happen if spin-transparent storage rings become adopted tools for such experiments. We present a new design of highly specialized table-sized storage rings, which use low-energy polarized electron beams and Mott polarimetry. Based on the spin transparency ansatz, the spin precession stemming from the magnetic dipole moment is canceled at any beam energy after an electron's turn along the periodic orbit in the ring. Meanwhile, a spin precession induced by the fundamental physics of interest, e.g., the electron's permanent electric dipole moment (EDM) and/or ultralight-dark-matter-mediated forces such as axions, will accumulate. However, capitalizing on such types of rings is not only desirable for measurements of EDMs and axion searches relevant to $CP$ violation and matter-antimatter asymmetry in the Universe, but may also find very promising applications in quantum computing.

nucl-ex↗
arXiv2026

Controlling Atom Array in an Ultra-high-cooperativity Optical Cavity

Neutral-atom array and cavity quantum electrodynamics offer complementary strengths for quantum science: scalable, reconfigurable qubit architectures and strong coherent light-matter coupling. Combining them in a single platform requires an optical cavity with simultaneously high cooperativity, sufficient mode volume to accommodate atom array, and ample side optical access for atom trapping, imaging, cooling, and rearrangement, a combination that is challenging to achieve. Here we realize an atomic array integrated with a millimeter-scale Fabry--P\'erot cavity whose optically-characterized single-atom cooperativity reaches $\eta_{\mathrm{cav}}=125\pm13$. Atom-cavity transmission spectra of trapped atoms yield an effective spectroscopic cooperativity $\eta_{\mathrm{spec}}=112.3\pm3.3$, providing an in-situ verification of strong coupling in the integrated platform, and we demonstrate simultaneous coupling of up to 16 individually trapped atoms to the antinode of the cavity mode. The key technical advance is a two-step mirror-fabrication method combining precision mechanical shaping and carbon-dioxide laser polishing, which produces concave fused-silica mirrors with sub-millimeter radii of curvature and residual roughness below 2 \r{A}. Our results establish a regime of cavity-integrated atomic array that simultaneously provides high cooperativity, large mode volume, and flexible manipulation of individual atoms, opening opportunities for cavity-assisted quantum state readout and long-range entanglement-engineering in atom-array platforms.

quant-ph↗
arXiv2026

Contaminated Multi-task Learning with Heterogeneity: Fundamental Limits and Optimal Algorithms

Integrating information across related tasks can improve estimation and prediction in transfer, multi-task, and federated learning, but contamination and heterogeneity make robust borrowing challenging. We study a contaminated multi-task empirical risk minimization (ERM) framework in which an $\epsilon$ fraction of $K$ tasks, each with sample size $n$, may be arbitrarily contaminated while the remaining tasks are heterogeneous. Our goal is to estimate both the global minimizer of the average risk and the clean task-specific minimizers, thereby combining robustness and personalization. In the Gaussian mean model, we show that several common paradigms, including adaptive and robust regularization around a shared center, global matrix regularization, decomposition-based regularization, and score-based outlier-task detection, all suffer from a worst-case contamination error of order $\epsilon\sqrt{d/n}$, which is suboptimal compared to the lower bound $\epsilon/\sqrt{n}$. This identifies a dimension-dependent barrier for these approaches. We then establish minimax lower bounds for a general heterogeneous ERM setting and propose a computationally efficient filtering-based robust multi-task gradient descent method. Under local strong convexity, smoothness, and sub-Gaussian gradient assumptions, the proposed method attains high-probability upper bounds matching the minimax rates up to logarithmic factors over a broad regime. In particular, it removes the extra $\sqrt{d}$ contamination dependence of many regularization-based methods and score-based outlier detection, while achieving personalization to local tasks under strong heterogeneity. Simulations and a real-data analysis demonstrate strong robustness and personalization relative to a broad range of benchmark methods.

stat.ML↗
arXiv2026

Plan Right, Then Plan Tight: Symbolic RL for Efficient Embodied Reasoning

Embodied task planning asks an agent to turn a natural-language instruction into an executable sequence of actions in a physical scene, and is a building block for household, assistive, and service robots. Recent prompting-based and reinforcement-learning planners generate fluent action text but lack a cheap deterministic check that the produced plan is valid in the target world, while high-fidelity simulation is too slow to serve as an inner-loop training signal. The general problem is therefore how to obtain verifiable supervision and rewards for embodied planners without relying on string-level matching or full simulation. Here we show that a single BDDL specification, automatically constructed from open-world video evidence or curated tasks, can serve as a shared interface for data construction, plan verification, and reward design. A video-to-BDDL parser, an LLM verifier, and a lightweight symbolic engine together supply dense feedback at millisecond latency. We further introduce GroupAdapt, a difficulty-aware length schedule that uses the in-batch group pass rate as a zero-cost signal so that hard prompts get wider length tolerance and automatically tighten as their pass rate improves. Under the guidance of the proposed verifier and GroupAdapt schedule, the 8B planner attains a Strict-Pass score of 97.3 on BEHAVIOR-1000, yielding a 25.9 percent relative improvement over the Qwen3-8B baseline. This result exceeds the strongest large-model baseline by 3.5 percent, while simultaneously compressing the response length by 79 percent to 207 tokens, demonstrating both effectiveness and efficiency.

cs.RO↗
arXiv2026

Multi-Source Transfer Learning of Sparse Single-Index Models

Transfer learning leverages knowledge from related source domains to improve learning in a target domain. Recent theoretical advances cover a broad range of regression settings within (generalized) linear models. Despite their diversity, these methods share two common constraints: they assume a known link function or linear structure and require direct access to raw source data. To move beyond these constraints, we propose a source-data-free transfer learning framework based on the single-index model (SIM). Instead of requiring raw source data, our method transfers only summary statistics derived from a generalized Stein's lemma in a one-time communication. This design preserves privacy and avoids side effects caused by dissimilarities of unknown nonlinear link functions across domains. To capture flexible, unknown nonlinearity, we employ a multilayer perceptron guided by the pre-estimated index from the transferred statistics, which significantly mitigates overfitting. Extensive experiments on synthetic data and a real-world application demonstrate consistent improvements over existing (generalized) linear model-based approaches. The proposed framework thus offers a practical, privacy-preserving, and nonlinear-adaptive solution for transfer learning.

stat.ME↗
arXiv2026

RGLD: Randomized Global-Local Density Estimation for Tabular Anomaly Detection

Unsupervised tabular anomaly detection requires methods that are accurate, robust across heterogeneous datasets, and computationally efficient. Classical statistical detectors are often efficient, but they usually rely on a fixed data view and a single notion of abnormality. Deep anomaly detectors can learn more flexible scoring functions, but they are substantially slower and difficult to tune in unsupervised settings due to the lack of a reliable supervisory signal. We propose RGLD, a randomized global-local density estimator for efficient unsupervised tabular anomaly detection. RGLD combines a global random-feature density branch, which identifies samples in broadly low-density regions, with a local neighbor branch, which detects samples that are weakly supported by nearby observations. Both branches operate over feature-bagged randomized views, allowing RGLD to expose anomaly evidence that may be hidden in any single representation. We conduct experiments on 47 tabular datasets against 23 statistical and deep anomaly detection baselines under fully unsupervised setting. RGLD achieves the strongest dataset-level AUROC performance, ranking 1st in dataset wins, and ranks 2nd in AUPRC wins. RGLD is also faster than all evaluated deep detectors, achieving 50x-580x speedups, and remains competitive with statistical methods in runtime, yielding a favorable accuracy-efficiency tradeoff.

cs.LG↗
arXiv2026

PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models

Multimodal large language models (MLLMs) have achieved remarkable progress in visual understanding tasks. However, most existing MLLMs rely on autoregressive generation, which limits their efficiency for perception tasks that require captioning multiple regions. In this work, we propose PerceptionDLM, a multimodal diffusion language model optimized for efficient parallel region perception. Built upon PerceptionDLM-Base, a strong foundational baseline that achieves state-of-the-art performance among open-source diffusion MLLMs, our architecture fully leverages the parallel decoding nature of DLMs. Specifically, we introduce efficient prompting and structured attention masking to enable simultaneous perception of multiple masked regions, allowing the model to generate region descriptions in parallel at both the sequence and token levels. This design significantly improves inference efficiency compared with existing approaches that process regions sequentially. To systematically evaluate the parallelism property of visual perception capability for DLMs, we construct a new Parallel Detailed Localized Captioning Benchmark (ParaDLC-Bench) by scaling the DLC-Bench to include multiple region masks per image, enabling joint evaluation of both caption quality and inference efficiency. Experiments demonstrate that PerceptionDLM maintains competitive performance in region captioning while achieving substantial speed improvements for multi-region perception tasks. Our results highlight the potential of multimodal diffusion language models for efficient, parallel visual perception. To the best of our knowledge, we are the first to achieve parallel region caption and perception by leveraging the advantages of diffusion language models. Code, models, and datasets are released.

cs.CV↗
arXiv2026

LoomVideo: Unifying Multimodal Inputs into Video Generation and Editing

Developing unified video generation and editing models capable of interpreting interleaved multimodal inputs is a promising yet challenging frontier field. Existing unified frameworks predominantly rely on massive models (typically 13B parameters or more) and incorporate source video conditions for editing by concatenating sequence tokens. This concatenation inevitably doubles the sequence length, quadrupling the computational complexity of the self-attention mechanism and introducing prohibitive overhead. To address these bottlenecks, we present LoomVideo, a highly efficient 5B-parameter unified architecture for both video generation and editing. LoomVideo replaces the standard text encoder with a Multimodal Large Language Model (MLLM) and employs Deepstack injection mechanism to align multi-layer MLLM features with the Diffusion Transformer (DiT). Crucially, we introduce a zero-overhead Scale-and-Add conditioning approach for video editing. By scaling and directly adding the clean source video latent to the noised target latent, this elegant design eliminates the need for token concatenation, drastically reducing computational cost while maintaining robust capabilities for complex, non-rigid edits. Furthermore, a Negative Temporal RoPE strategy is seamlessly integrated to handle multiple reference images. Extensive experiments demonstrate that our compact 5B model achieves state-of-the-art or highly competitive performance across comprehensive benchmarks, exhibiting exceptional superiority in e-commerce and fashion generation scenarios. Benefiting from the zero-overhead conditioning mechanism, LoomVideo achieves at least a 5.41x acceleration in inference speed compared to models of similar capabilities, paving the way for highly practical and efficient video foundation models.

cs.CV↗
arXiv2026

Trapping 11,000 Atoms in a Tweezer Array Generated by a Single Metasurface

The scalability of physical qubit numbers is a central challenge toward a universal fault-tolerant quantum computer. The inherent scalability of atom array quantum computers stems from the identical nature of atomic qubits, so the available qubit resource is primarily limited by the number of atoms that can be trapped and controlled. Here, we robustly trap 11,000 individual atoms in a tweezer array, thereby enabling the available qubit resource to reach the tens-of-thousands scale for the first time among all quantum computation platforms. This advance is enabled by a single metasurface, approximately 2 cm in diameter, that generates the entire tweezer array without the need for microscope objectives, thereby maximizing laser-power efficiency. The large aperture ensures a working distance of about 1.5 cm, allowing the metasurface to be placed outside the vacuum cell and avoiding the technical complications of in-vacuum operation. We further characterize the randomly loaded atom array using the statistical theory of percolation phase transitions. This work takes an important first step toward a quantum computer at the 10,000-qubit scale.

quant-ph↗
arXiv2026

Hyperedge approximation for stochastic processes on higher-order networks

Graphs are a standard framework for describing dynamical processes shaped by pairwise interactions among agents. But many systems involve interactions in groups of three or more agents. Here, we develop a method of "$\ell$-hyperedge approximation", a framework to analyze stochastic population processes on regular hypergraphs, in which each individual belongs to $k$ groups of size $\ell$. The framework accommodates both higher-order interactions that determine payoffs and higher-order processes for updating states in response to payoffs. Applied to evolutionary game dynamics, the framework generalizes the classical pairwise result on benefits and costs, $b/c>k$, that favors the spread of cooperation; and it provides critical benefit-to-cost ratios for nonlinear $\ell$-player public goods games that cannot be reduced to pairwise interactions. Applied to complex contagions, where inheritance of states occurs within hyperedges rather than along parent-offspring edges, the framework gives a closed-form result for the fixation probability, which shows how a complexity parameter governs the spread of rare types. Coupling the two processes produces a single stochastic model of payoff-biased complex contagion in structured populations. These results extend pair approximation from graphs to hypergraphs, accommodating multi-way interactions and inheritance structures with no pairwise analog.

physics.soc-ph↗
arXiv2026

Sparse Fluid Antenna Arrays: Continuous Position Design Beyond Classical DOF Limits

Fluid antenna system (FAS), which continuously repositions a single physical element across a deployment region $[0, D]$, breaks this limit by freeing antenna positions from the discrete grid entirely. This paper establishes the theoretical foundations of sparse FAS design for direction-of-arrival (DOA) estimation and shows that continuous position freedom unlocks three compounding advantages over the classical designs. \emph{First}, we derive a universal dual DOF bound and prove that FAS-optimized positions can approach it, growing the DOF linearly with $D/\lambda$ , where $\lambda$ is the signal wavelength, rather than saturating at $O(N^2)$. \emph{Second}, the CRB scales as $O(1/D^{2L})$ for $L$ sources, a $(D/(N^2 d_0))^{2L}$ improvement over the best grid design, with $d_0 = \lambda/2$ and D-optimal positions admitting closed-form solution for single sources and efficient Frank-Wolfe algorithm for multiple sources. \emph{Third}, we propose a two-stage FAS-MUSIC approach that combines coarray MUSIC disambiguation with full-aperture local maximum likelihood (ML) refinement to track the CRB, overcoming the grating-lobe ambiguity inherent in large-aperture non-uniform arrays. Robustness to minimum spacing constraints, mutual coupling, and finite position accuracy is also analyzed. Extensive simulations show that FAS-MUSIC achieves $17.5\times$ lower root mean squared error (RMSE) than uniform linear array (ULA) MUSIC and that FAS with $4$ antennas outperforms MRA with $8$ antennas, gains that are unattainable by any grid-constrained design.

eess.SP↗
arXiv2026

Benchmarking and Evolving Reason-Reflect-Rectify for Reflective Visual Generation

Text-to-Image (T2I) models and Unified Multimodal Models (UMMs) have achieved remarkable progress in visual generation. However, their reliance on a single-pass generation paradigm limits their ability to handle complex prompts requiring iterative refinement. To enable multi-round Reflective Visual Generation (RVG), we formalize the Reason-Reflect-Rectify (R^3) loop as a core framework and introduce R^3-Bench, a benchmark of over 600 expert-annotated instances that quantifies iterative reasoning and rectification capabilities. Evaluation on R^3-Bench reveals a critical gap: while state-of-the-art models can identify generation errors, they fail to generate actionable rectification instructions. To bridge this gap, we propose R^3-Refiner, a dual-stage framework leveraging Group Relative Policy Optimization (GRPO) and a Hierarchical Reward Mechanism (HRM) to better align rectification with reflective reasoning. Experiments show that R^3-Refiner achieves significant improvements on R^3-Bench (+12.0% in Reflective Verdict Score, +9.0% in Rectification Score), and can be seamlessly integrated with various MLLMs to enhance the generation quality of different T2I models on GenEval++ and T2I-CompBench. Code is available at https://github.com/xiaomoguhz/R3-Bench.

cs.CV↗
arXiv2026

Large Vision-Language Models Get Lost in Attention

Despite the rapid evolution of training paradigms, the decoder backbone of large vision--language models (LVLMs) remains fundamentally rooted in the residual-connection Transformer architecture. Therefore, deciphering the distinct roles of internal modules is critical for understanding model mechanics and guiding architectural optimization. While prior statistical approaches have provided valuable attribution-based insights, they often lack a unified theoretical basis. To bridge this gap, we propose a unified framework grounded in information theory and geometry to quantify the geometric and entropic nature of residual updates. Applying this unified framework reveals a fundamental functional decoupling: Attention acts as a subspace-preserving operator focused on reconfiguration, whereas FFNs serve as subspace-expanding operators driving semantic innovation. Strikingly, further experiments demonstrate that replacing learned attention weights with predefined values (e.g., Gaussian noise) yields comparable or even superior performance across a majority of datasets relative to vanilla models. These results expose severe misallocation and redundancy in current mechanisms, suggesting that state-of-the-art LVLMs effectively ``get lost in attention'' rather than efficiently leveraging visual context.

cs.AI↗