arXiv subjects

Hao Wu

Publications and source records attributed to Hao Wu.

At least 163 records · Page 9Linked to original sources

arXiv2026

ActProbe: Action-Space Probe for Early Failure Detection of Generative Robot Policies

Generative robot policies fail unpredictably at deployment: they hesitate at critical moments, drift off-task, or commit to unrecoverable actions. Existing online failure detectors either require white-box access to policy internals or add runtime overhead through resampling and observation-side signals. Our empirical analysis shows that emitted action chunks themselves already carry strong predictive signal for impending failures in generative robot policies. Motivated by this observation, we introduce ActProbe, a lightweight, pure action-space detector that uses two compact signals available from a single forward pass: Temporal Consistency Error (TCE) between consecutive action chunks and Action Chunk Magnitude (ACM) of the current chunk. ActProbe maps these signals to per-step failure probabilities with a task-conditioned LSTM-MLP architecture. Across a diverse suite of generative robot policies and benchmarks, ActProbe raises alerts before failures become visually recognizable, improving the accuracy (F1)-timeliness Pareto frontier of failure detection by an average hypervolume gain of +12.7% over both internal- and external-feature baselines, with a +9.0% early-detection ROC-AUC lead on unseen tasks. ActProbe further transfers to deployment, predicting failures on unseen real-robot pick tasks and accelerating RL fine-tuning (PPO) with 2.9x fewer environment interactions.

cs.RO↗

arXiv2026

Bypassing Copyright Protection in Diffusion-based Customization via Two-Stage Latent Feature Optimization

With the growing concerns over copyright infringement in diffusion-based customization, adversarial attacks have emerged as a prominent defense strategy to prevent malicious content forgery in personalized image generation. However, current defenses typically introduce persistent perturbations in the latent space of Latent Diffusion Models (LDMs), which remain susceptible to adaptive bypasses by adversaries. In this paper, we introduce Two-Stage Latent Feature Optimization (TS-LFO), an efficient and effective copyright-stealing attack against protected diffusion-based customization. We begin by observing that existing defenses primarily disrupt the mapping between input images and their latent representations, thereby degrading the model's ability to produce personalized outputs. To counteract this, TS-LFO restores the broken mapping through a two-stage optimization process. In the Latent Denoising Stage, we enhance semantic consistency between latent codes and input images by jointly minimizing a Latent-Image Alignment Loss and a Latent Diffusion Loss with timestep-dependent weights, effectively suppressing the high-frequency noise introduced by defenses. In the Latent Reconstruction Stage, we recover low-frequency semantic information using pixel-level constraints to refine the latent features. Extensive experiments show that TS-LFO consistently bypasses state-of-the-art (SOTA) copyright defenses and outperforms SOTA copyright attacks such as DiffPure, GrIDPure and IMPRESS across diverse settings.

cs.CR↗

arXiv2026

Almost Sure Bounds for Discrepancies of Linear Forms on the Circle

As a generalization of irrational rotations and a dual case of higher-dimensional Kronecker sequences, we study the discrepancy of sequences of linear forms on the circle. Given irrationals $α_1,\dots,α_d$, consider the set of $N^d$ points $\{k_1α_1+\cdots+k_dα_d \mod 1 : 1\le k_j\le N\}$. We prove that for a full-measure set of vectors $(α_1,\dots,α_d)\in\mathbb{R}^d$, the maximal discrepancy of these points relative to intervals in $[0,1)$ has the optimal principal order $(\log N)^d$, up to powers of $\log\log N$. This result provides a nearly sharp dual analogue, in the setting of linear forms, to Beck's celebrated theorem on multidimensional Kronecker sequences (Ann. of Math., 1994). The proof combines Fourier analysis, metric multiplicative Diophantine estimates, and a duality argument which reduces certain lattice-counting errors to Beck's discrepancy theorem.

math.NT↗

arXiv2026

High-Speed Multi-Dimensional Optical Field Measurement via MMF-MCF Spatial-Temporal Mapping Architecture

Wavelength and state of polarization constitute fundamental dimensions of optical fields. While simultaneous quantification of these parameters is critical, existing methodologies often lack the speed required for real-time analysis. Here, we present a compact high-dimensional optical field analyzer employing a discrete spatiotemporal sampling architecture based on multimode and multicore fibers. An optical delay line array maps spatial speckle patterns into serial pulse sequences and facilitates efficient single-pixel detection. Leveraging a residual multilayer perceptron network, the system attains a wavelength mean absolute error of 0.25 pm and a polarization resolution of 0.2015 (in normalized Stokes space). Analysis of the spatial sampling density reveals that 5-6 sampling points are required to balance measurement rate and accuracy. Notably, the system exhibits isotropic fault tolerance against single-core failures. This confirms that optical field information is redundantly encoded across the entire fiber cross-section rather than localized in specific channels. This framework provides a solution for multiparameter decoupling under severe spatial downsampling and useful insights for the design of next generation high-speed and robust all-fiber analysis systems.

physics.optics↗

arXiv2026

Struct-Searcher: Agentic Structural Thinking Advances Multimodal Deep Information Seeking

Deep research agents have attracted increasing attention for their ability to collect large-scale online information to acquire target knowledge, with recent efforts shifting from purely text-based information seeking to multimodal settings. However, existing agentic workflows are largely aligned with evidence accumulation models, which linearly aggregate evidence and lack principled mechanisms for handling contradictory information across heterogeneous modalities. Towards this end, we propose Struct-Searcher, a structural agentic workflow grounded in belief revision theory that explicitly maintains an evolving multimodal structural graph throughout the reasoning process, enabling effective conflict-aware multimodal deep information seeking. Extensive experiments across multiple benchmark datasets and backbone models demonstrate that Struct-Searcher is (1) plug-and-play and model-agnostic, yielding an average relative accuracy improvement of 17.2% on BrowseComp-VL across five different backbones. (2) top-performing, consistently outperforming state-of-the-art vision-language models (VLMs) and deep research agents, with relative accuracy improvements of 3.7% on MM-BrowseComp, 1.5% on HLE-VL, and 0.7% on BrowseComp-VL over the second-best competing approach.

cs.CV↗

arXiv2026

vLLM Semantic Router: Signal Driven Decision Routing for Mixture-of-Modality Models

As large language models (LLMs) diversify across modalities, capabilities, and cost profiles, the problem of intelligent request routing: selecting the right model for each query at inference time, has become a critical systems challenge. We present vLLM Semantic Router, a signal-driven decision routing framework for Mixture-of-Modality (MoM) model deployments. The architecture follows two complementary Shannon-inspired views. In the information-theoretic regime, signal extraction reduces the entropy of "which model?" by distilling routing-relevant information from raw queries. In the Boolean-algebraic regime, the decision engine composes functionally complete routing policies from signal conditions. The central innovation is composable signal orchestration: thirteen heterogeneous signal types, spanning sub-millisecond heuristics and neural classifiers for semantics, safety, and modality, are composed through configurable Boolean decision rules into deployment-specific routing policies, so that fundamentally different scenarios (multi-cloud enterprise, privacy-regulated, cost-optimized) are expressed as different configurations over the same architecture. Matched decisions drive semantic model routing via thirteen selection algorithms, while per-decision plugin chains enforce safety constraints including a three-stage HaluGate hallucination detection pipeline and a lightweight episodic memory system with ReflectionGate for personalized multi-turn context. A typed neural-symbolic DSL specifies these routing policies and compiles them to multiple deployment targets, enabling configuration-first adaptation without code changes. Together, these components show that composable signal orchestration enables a single framework to serve diverse deployment scenarios with differentiated cost, privacy, and safety policies.

cs.NI↗

arXiv2026

Extended MF-FBSDEs with nonlinear domination-monotonicity conditions and stochastic optimal controls of Linear System with quadruple controls

This paper extends the domination-monotonicity conditions, which guarantee the well-posedness of extended mean-filed forward-backward stochastic differential equations (extended MF-FBSDEs), from the previously studied linear framework to a nonlinear setting by incorporating nonlinear adjoint functions. Utilizing this generalized well-posedness result for extended MF-FBSDEs in conjunction with other refined analytical techniques, we address two classes of stochastic quadruple optimal controlled problems: a linear-convex problem and a linear-quadratic problem with input constraints that are permitted to be time-dependent and random. For each problem, we establish the existence and uniqueness of optimal controls and derive their explicit closed-form representations.

math.OC↗

arXiv2026

PnP-Corrector: A Universal Correction Framework for Coupled Spatiotemporal Forecasting

Coupled spatiotemporal forecasting is important for predicting the future evolution of multiple interacting dynamical systems, such as in climate models. However, existing methods are severely constrained by the persistent bottleneck of compounding errors. In coupled systems, errors from each subsystem simulator propagate and amplify one another, a phenomenon we term Reciprocal Error Amplification, leading to a rapid collapse of long-range predictions. To address this challenge, we propose a universal framework called PnP-Corrector (Plug-and-Play Corrector). The core idea of our framework is to decouple the physical simulation from the error correction process: it freezes pre-trained physics simulation engines and exclusively trains a correction agent to proactively counteract the systematic biases emerging from the coupled system. Furthermore, we design an efficient predictive model architecture, DSLCast, to serve as the backbone of this framework. Extensive experiments demonstrate that our method significantly enhances the long-term stability and accuracy of coupled forecasting systems. For instance, in the challenging task of a 300-day global ocean-atmosphere coupled forecast, our PnP-Corrector framework reduces the prediction error of the baseline model by 28% and surpasses state-of-the-art models on several key metrics.

cs.AI↗

arXiv2026

Tripod in uniform spanning tree and three-sided radial SLE$_2$

Fix a bounded $3$-polygon $(Ω; x_1, x_2, x_3)$ with three marked boundary points $x_1, x_2, x_3\in\partialΩ$ and suppose $(Ω^δ; x_1^δ, x_2^δ, x_3^δ)$ is an approximation of $(Ω; x_1, x_2, x_3)$ on $δ$-scaled hexagonal lattice. We consider uniform spanning tree (UST) in $Ω^δ$ with wired boundary conditions. Conditional on the event that both branches from $x_1^δ$ and $x_2^δ$ hit the boundary through $x_3^δ$, the two branches meet at a point $\trifurcation^δ$ which we call trifurcation, and the union of the three branches from $x_j^δ$ to $\trifurcation^δ$ form a tripod in the UST. We compute the scaling limit of the tripod: the distribution of trifurcation is absolutely continuous with respect to Lebesgue measure with explicit density; given the trifurcation, the conditional law of the tripod is three-sided radial SLE$_2$. The proof relies on construction of a new observable for trifurcation in our key lemma--Lemma~3.1--where we use Fomin's formula and the geometry of the hexagonal lattice in an essential way. Interestingly, the scaling limit of the observable for trifurcation coincides with the partition function for three-sided radial $\SLE_2$. Our result gives a probabilistic interpretation of the correlation function in CFT which has conformal weights $1$ at the three boundary points and has a spinless field of weights $(1,1)$ at the bulk point.

math.PR↗

arXiv2026

PRISM: Position-encoded Regressive Inverse Spectral Model for Multilayer Thin-Film Design

The inverse problem of multilayer thin-film optical coatings design represents a complex combinatorial-continuous optimization challenge. We present PRISM (Position-encoded Regressive Inverse Spectral Model), a unified decoder-only autoregressive transformer that streamlines this process by jointly predicting discrete material selection and continuous thickness regression within a single backbone. PRISM introduces two primary architectural innovations: (1) spectrum prefix conditioning, which utilizes standard prefix tokens for in-context target injection, and (2) cumulative-depth Rotary Position Embeddings, which encode continuous thickness directly into the positional representation to preserve the physical spatial relationships of the stack. Our benchmarks demonstrate that a PRISM-13M model reduces MAE by over 50\% compared to other transformer baselines while utilizing only one-fifth of the parameters. Furthermore, a 44M-parameter variant achieves state-of-the-art performance (MAE = 0.010) on our in-distribution validation benchmark and operates significantly faster than simulated annealing, offering a highly efficient alternative to classical optimization methods.

cs.LG↗

arXiv2026

Modeling and Optimization for Massive Data Allocation in Database

In the era of big data, e-commerce and Internet platforms face the challenge of processing massive amounts of data. However, due to data being scattered across different machines in distributed database, extra communication costs are incurred in gathering relevant data to complete transactions. Without a carefully designed data placement scheme, this cost can severely impact the performance of Online Transaction Processing systems. To meet industry requirements, algorithms that output a data placement scheme that achieves i) data balance and ii) low communication overhead within a fixed period of time are eagerly investigated. Although some existing methods have been studied, they do not adequately meet the aforementioned requirements. In this paper, inspired by the normalized cut of spectral clustering, we introduce a novel model for data allocation problem. The normalized cut reconciles the inherent conflict between the two objectives. Taking into account the variable characteristics of the model, we formulate the problem as a 0-1 optimization problem, and solve the relaxed problem using the Bregman proximal gradient method with guaranteed convergence. The numerical experiments reveal that the convergent solutions can be smoothly rounded to discrete solutions. Furthermore, our algorithm surpasses both simple and meta-heuristic partitioning schemes by minimizing migration costs while maintaining a superior balance.

cs.DB↗

arXiv2026

Odd spin symmetry and anisotropy switching in p-wave magnet CeNiAsO

Odd-parity magnets, complementary to altermagnets, exhibit unique properties such as high efficiency in charge-spin conversion and compatibility with conventional superconductivity, of critical importance in the pursuit of energy-efficient spintronics and topological superconductors for quantum computation. For even-parity d-wave and g-wave altermagnets, the magnetic structure, spin-split band structure and physical properties are currently under intensive study. On the contrary, while hundreds of odd-parity magnets and the promising properties have been predicted in theory, experimental studies are scarce. Specifically, the magnetic structure and transport properties of candidates NiI2 and Ga3Ru4Al12 have been reported, yet the characteristic band structure and particularly the odd-parity spin symmetry remain elusive. Here we demonstrate experimentally the deterministic p-wave spin symmetry and resistance anisotropy switching for the prototype odd-parity magnet, CeNiAsO. Angle-resolved photoemission spectroscopy (ARPES) reveals two cleaved terminations with distinct surface band structure. By compensating the polar surface, we achieve intrinsic bulk band structure, for which the spin splitting can be well described by the p-wave magnetic structure through first-principles calculation. The bulk spin polarization measured by spin-resolved ARPES exhibits symmetry with only one degenerate plane, fingerprint of p-wave magnetism. We further demonstrate giant resistance anisotropy and switching between high-resistance and low-resistance states through modest field-induced domain selection, highlighting its potential for antiferromagnetic spin memory devices. The structural similarity between CeNiAsO and 1111-type Fe-based superconductors stimulates further exploration on the interplay between p-wave magnetism, superconductivity and band topology.

cond-mat.str-el↗

arXiv2026

IP-Adapter Is All You Need: Towards Fine-Tuning-Free Diffusion-Based Talking Face Generation

With the rapid advancement of diffusion models, talking face generation has made remarkable progress. However, existing diffusion-based methods still require task-specific fine-tuning and large-scale audiovisual datasets, resulting in high computational costs that hinder scalability and accessibility of diffusion-based approaches across the research community. To address this, we propose a finetuning-free paradigm that directly performs talking face generation using the pretrained weights of Stable Diffusion and IP-Adapter. This backbone leverages the visual embedding capability of IP-Adapter to mine lip-related semantics from the pretrained Stable Diffusion. To address the challenges of identity drift, synchronization errors, and temporal instability, we also design three trainable-parameterfree components: (1) the Structurist, which explicitly disentangles and reassembles lip and appearance features to mitigate identity drift and appearance distortion; (2) the Structure Controller, which adaptively refines embeddings based on quasi-monotonic motion trends for precise lip synchronization; and (3) the Noise Sensor, which introduces Gaussian prior to detect and suppress flicker and jitter artifacts and enhance temporal consistency. Experimental results show that our method outperforms existing SOTA approaches in both lip-sync accuracy (at least 0.16 gain in PCLD) and visual fidelity (at least 0.7 improvement in FID), establishing a novel fine-tuning-free diffusion framework for talking face generation.

cs.CV↗

arXiv2026

ViCA: Efficient Multimodal LLMs with Vision-Only Cross-Attention

Modern multimodal large language models (MLLMs) adopt a unified self-attention design that processes visual and textual tokens at every Transformer layer, incurring substantial computational overhead. In this work, we revisit the necessity of such dense visual processing and show that projected visual embeddings are already well-aligned with the language space, while effective vision-language interaction occurs in only a small subset of layers. Based on these insights, we propose ViCA (Vision-only Cross-Attention), a minimal MLLM architecture in which visual tokens bypass all self-attention and feed-forward layers, interacting with text solely through sparse cross-attention at selected layers. Extensive evaluations across three MLLM backbones, nine multimodal benchmarks, and 26 pruning-based baselines show that ViCA preserves 98% of baseline accuracy while reducing visual-side computation to 4%, consistently achieving superior performance-efficiency trade-offs. Moreover, ViCA provides a regular, hardware-friendly inference pipeline that yields over 3.5x speedup in single-batch inference and over 10x speedup in multi-batch inference, reducing visual grounding to near-zero overhead compared with text-only LLMs. It is also orthogonal to token pruning methods and can be seamlessly combined for further efficiency gains. Our code is available at https://github.com/EIT-NLP/ViCA.

cs.CV↗

arXiv2026

Spin-Hall-Like Magnon Transport in a Synthetic Antiferromagnetic Skyrmion Lattice

We investigate spin-Hall-like magnon edge transport in a synthetic antiferromagnetic skyrmion lattice composed of two antiferromagnetically coupled skyrmion lattice layers with opposite magnetic textures. Based on a relaxed bilayer texture from micromagnetic simulations, we construct the bosonic Bogoliubov-de Gennes Hamiltonian within linear spin-wave theory and calculate the bulk and strip magnon spectrum. We find counterpropagating in-gap edge modes with opposite layer polarization, whose layer-resolved propagation is further confirmed by dynamical micromagnetic simulations. A symmetry analysis shows that the fully coupled system lacks the pseudo-time-reversal symmetry required for a genuine bosonic Z2 topological phase. Thus, the observed edge modes are not Z2-protected helical magnon edge states, but layer-polarized, spin-Hall-like modes originating from the opposite Hall tendencies of the two skyrmion lattice layers. These results establish synthetic antiferromagnetic skyrmion lattices as a platform for spin-Hall-like magnon transport beyond a strict bosonic Z2 classification.

cond-mat.mes-hall↗

arXiv2026

Reasoning Matters: Mitigate Hallucination in Multimodal Large Reasoning Models via Reasoning-Conditioned Preference Optimization

Multimodal Large Reasoning Models introduce the reasoning paradigm, demonstrating strong capabilities on complex vision-language tasks. However, they still suffer from severe hallucinations. Existing training-based methods typically mitigate hallucinations through response-level direct preference optimization (DPO), where the Chain-of-Thought (CoT) and the final answer are treated as a monolithic output and optimized jointly. We reveal that this formulation performs similarly to answer-only optimization, suggesting that it primarily learns answer-level preference, while leaving CoT-level supervision insufficiently exploited. To address this issue, we explicitly formulate a CoT-oriented preference term and derive Reasoning-Conditioned Direct Preference Optimization (RC-DPO), which models the CoT as a condition for answer generation and contrasts the preference for the same preferred answer under different CoT conditions, promoting answer-supportive reasoning chain alignment. To further improve optimization, we introduce a reasoning-enhanced preference data generation strategy that employs Monte Carlo Tree Search to discover visually grounded and logically consistent CoTs as positive samples, and attention-guided CoT token pruning to construct negative ones. Extensive experiments across various models and benchmarks show that RC-DPO effectively mitigates hallucinations and improves the reliability of the multimodal reasoning process.

cs.AI↗

arXiv2026

FINOM: Fast Sinkhorn on Non-uniform Meshes

A linear-complexity algorithm for computing the Wasserstein-1 distance on non-uniform meshes is proposed. This work extends the fast Sinkhorn algorithms from [Q. Liao et al., Commun. Math. Sci., 20(2022)] and [Q. Liao et al., J. Sci. Comput., 98 (2024)] to non-uniform meshes. In those prior works, a distinctive collinear structure of the kernel matrix on uniform meshes was identified, enabling $O(N)$ acceleration via dynamic programming. While non-uniform meshes are prevalent in practical applications like computational fluid dynamics and finance, their lack of collinearity has hindered direct acceleration. In this paper, we introduce the concept of a ``dividing index'', which partitions the kernel matrix into two blocks. We demonstrate that each block exhibits a quasi-collinear property, a generalization of the structure found in uniform meshes. Leveraging this insight, we develop \textbf{F}ast S\textbf{I}nkhorn algorithm on \textbf{NO}n-uniform \textbf{M}eshes (\textbf{FINOM}), a dynamic programming approach that reduces the per-iteration complexity of the Sinkhorn algorithm from $O(N^2)$ to $O(N)$. Extensive numerical experiments on 1D and 2D problems confirm these improvements, achieving speed-ups of several orders of magnitude while maintaining accuracy.

math.NA↗

arXiv2026

Heterogeneous Parallelism for Multimodal Large Language Model Training

Foundation model training is becoming multimodal, from post-training pipelines to large-scale pretraining. As modality coverage broadens, context windows grow, and encoder LLM scales diverge, a single LLM-centric TP/CP/PP/DP/EP layout increasingly limits throughput. This coupling forces encoders to inherit LLM-driven sharding and placement choices that can add communication, limit encoder parallelism, or constrain the LLM schedule; the mismatch is most pronounced at long contexts, where LLM context parallelism is needed for the fused multimodal sequence but encoder inputs remain bounded. We present heterogeneous parallelism for multimodal large language model training, an abstraction that lets modules in one end-to-end graph use independent layouts and rank placements, supporting colocated execution on shared GPUs and non-colocated execution on disjoint rank sets. The key challenge is preserving boundary tensor semantics across independent layouts: forward activations must be materialized for the destination layout, while backward gradients must be routed back to the source layout. We address this with boundary communicators that implement forward and backward layout transforms, plus scheduling extensions for both placement modes. We evaluate optimized homogeneous, colocated heterogeneous, and non-colocated heterogeneous configurations across multimodal workloads and GPU scales to characterize when added layout and placement freedom exposes a better operating point. Across this sweep, colocated heterogeneity improves TFLOPS/GPU by up to 49.3%, while non-colocated heterogeneity improves aggregate token throughput by up to 13.0% and TFLOPS/GPU by up to 9.6%. We validate loss convergence parity against homogeneous baselines and release the system as an open-source Megatron-LM extension.

cs.LG↗