arXiv subjects

Yan Wu

Publications and source records attributed to Yan Wu.

At least 19 recordsLinked to original sources

arXiv2026

ADAPT: Agile Diffusion Action Priors for Robust and Steerable Online Text-Driven Humanoid Control

We present ADAPT, an end-to-end framework for interactive, text-conditioned humanoid whole-body control. Unlike dominant text-to-motion pipelines that generate kinematic motions for a separate tracker, ADAPT solves language control with an end-to-end closed-loop control framework, where the robot must continuously respond to changing commands while maintaining balance, natural motion, and smooth transitions. ADAPT learns a diffusion-based action prior from text-labeled humanoid state-action trajectories, enabling diverse motion skills to be directly executed from language commands. To improve long-horizon robustness and smooth prompt switching, we train a lightweight residual reinforcement learning policy on top of the frozen diffusion controller. We further show that the same diffusion policy can be reused as a steerable text-conditioned motion prior for downstream task adaptation. Experiments demonstrate robust language-grounded skill execution, smooth interactive transitions, and style-preserving downstream control.

cs.RO↗

arXiv2026

Statistical Characterization and Block-EM Estimation of Frequency-Domain NSI for OFDM Systems in Bursty Impulsive Noise

Impulsive noise (IN), characterized by its high power and non-Gaussian distribution, poses a critical challenge in modern orthogonal frequency-division multiplexing (OFDM) systems, driven by the proliferation of electronic devices. Current IN mitigation techniques rely heavily on time-domain processing. These methods are applied before the discrete Fourier transform (DFT), introducing additional complexity, failing to align with OFDM's inherent frequency-domain processing flow, and risking the destruction of subcarrier orthogonality due to imperfect IN subtraction. To address these limitations, we propose a frequency-domain, block-based framework for mitigating IN. The statistical representation of IN in the frequency domain is first derived using a transformed Gaussian mixture model. Based on this model, we develop an optimal receiver that leverages perfect noise state information (NSI), thereby identifying scenarios in which NSI is critical. We then propose an unsupervised block-based expectation-maximization (EM) framework for NSI estimation and develop three variants for evaluation. These include a simple symbol-by-symbol variance-updated EM, a sequence-based transition-updated EM, and a MAP-based EM that exploits a sparsity-promoting prior to automatically prune the number of states. Our frequency-domain design operates after the DFT, seamlessly integrates with the OFDM processing chain, preserves subcarrier orthogonality, and leverages the known IN block structure to achieve substantial performance gains without the immense complexity of time-domain impulse reconstruction.

eess.SP↗

arXiv2026

Scaling Domain Data Repetition in LLM Pretraining

As large language models scale, their training-token budgets must also increase to maintain an appropriate tokens-per-parameter ratio ($\mathrm{TPP}$). However, high-quality domain data is much harder to scale than general web data. As model size and the training-token budget increase, its fraction in the training mixture tends to decrease. Repeating the available high-quality data provides an effective way to counteract this dilution, but excessive repetition may lead to overfitting. We study this trade-off under practical LLM scaling, where the training-token budget grows proportionally with model size. For a fixed domain, we first find that, surprisingly at a fixed $\mathrm{TPP}$, the optimal repetition count mildly increases with model size. Across different domains, we find that the optimal repetition count is strongly negatively correlated with the final validation loss of a domain: domains with lower loss can generally benefit from more repetitions. In contrast, the amount of unique domain data is only weakly related to the optimal repetition count. These findings suggest that repetition counts tuned on smaller proxy models with the same $\mathrm{TPP}$ can provide a practical estimate for larger models.

cs.AI↗

arXiv2026

MANGO-Grasp: Mahalanobis Fields over Geometry-Oriented 3D Gaussians for Cross-Embodiment Dexterous Grasping

Cross-embodiment dexterous grasping aims to synthesize stable grasps across heterogeneous multi-fingered hands with little or no embodiment-specific tuning. Existing interaction-centric methods achieve promising results, but their object representations often underrepresent local surface geometry, while their robot descriptors do not explicitly encode both robot morphology and kinematics. We propose MANGO-Grasp, an anisotropic interaction framework that represents objects as geometry-oriented 3D Gaussian primitives and robot hands as surface keypoints encoded into morpho-kinematic descriptors. The object primitives are adaptively allocated by geometric complexity and shaped as surface-aligned plates with outward normals, encoding local geometry. Mahalanobis fields over keypoint--primitive pairs serve as interaction prediction targets during training and as optimization guidance for grasp realization at inference. These fields rise sharply for displacement along the surface normal but only gently within the tangent plane, matching the directional structure of contact. Grasps are realized with one shared optimization formulation and hyperparameter setting across all embodiments. On the CMAP and MultiGripperGrasp benchmarks, MANGO-Grasp outperforms the strongest seen-hand baseline by up to 8.24 percentage points in simulation. It also transfers zero-shot to the unseen SharpaWave hand, improving over the strongest zero-shot baseline by up to 16.57 percentage points, and achieves 86% success in real-world experiments. The code and additional materials will be made available upon publication at https://connor-zh.github.io/MANGO-Grasp/.

cs.RO↗

arXiv2026

VistaVLA: Geometry- and Semantic-Aware 3D Gaussian-Grounded VLA for Robotic Manipulation

Vision-Language-Action (VLA) models have emerged as a powerful end-to-end paradigm for robotic manipulation by mapping language instructions and 2D visual inputs directly to actions. However, these models lack an explicit, scene-level 3D representation, limiting their ability to reason over spatial layouts and geometric constraints. While recent efforts incorporate explicit 3D cues, such as depth maps or point clouds, to improve geometric awareness, they primarily capture low-level structures and lack high-level semantic grounding in 3D space. In human cognition, interaction with the physical world relies on a 3D semantic cognitive map - an internal mental model that integrates spatial layouts with semantic context to enable persistent, viewpoint-invariant reasoning. In light of this, we present VistaVLA, a novel two-stage framework that constructs a geometry- and semantics-aware 3D cognitive representation from 3D Gaussian primitives and grounds it as compact context tokens for VLA policy learning. Specifically, VistaVLA lifts multi-view vision-language features into 3D Gaussian primitives, forming geometry-anchored semantic tokens that align view-consistent spatial grounding with 2D visual feature spaces. To make this 3D representation computationally tractable for effective VLA control, we introduce Merge-then-Query (MtQ), a token summarization mechanism. MtQ compresses dense Gaussian primitives into a highly compact set of spatially informative tokens, achieving a 99% token reduction while preserving action-relevant 3D layouts and semantic context. Extensive evaluations in both simulated and real-world environments demonstrate the effectiveness of VistaVLA. Notably, in real-world scenarios, VistaVLA improves success rates by 22.8% across seven real-world tasks and by 30.0% over the VLA-Adapter baseline on challenging out-of-distribution tasks.

cs.RO↗

arXiv2026

EAGOR: Embodied Reasoning in Omni-direction

Omni-directional (360{\deg}) cameras provide embodied agents with a holistic view of their surroundings, making them suited for directional reasoning in tasks such as navigation and object search. Existing Vision Language Models (VLMs) project 360{\deg} observations to 2D equirectangular projection (ERP) images and process them using architectures designed for perspective images. However, they ignore the spherical nature of 360{\deg} observations, where each pixel represents a viewing direction relative to the agent. Consequently, their direction estimates often become inconsistent under camera view transformations caused by agent motion. This limitation is particularly critical for map-free navigation, where the agent must continuously estimate the target direction in its egocentric frame. We propose EAGOR, a training-free, geometry-aware framework for embodied 360{\deg} directional reasoning. Instead of predicting target directions as ERP image coordinates, EAGOR formulates directional reasoning as recursive Bayesian estimation directly on the sphere. It maintains a continuous belief over target directions and propagates it equivariantly under agent motion without training the backbone VLMs. To achieve this, we introduce the Spherical Harmonic Belief Field (SH-BF), whose spherical harmonic representation provides a globally defined, rotation-aware basis for directional estimation on the spherical manifold. This formulation eliminates ERP seam discontinuities, latitude distortions, and interpolation errors. We evaluate EAGOR on two benchmark datasets and real-world experiments with a legged robot across directional reasoning tasks. EAGOR consistently outperforms existing methods, achieving average relative gains of +34.4% and +45.6% on HOS and OSR-Bench, respectively, while improving navigation success by +14.6%, reducing step count by 17.7%, and lowering mean angular error by 24.5%.

cs.RO↗

arXiv2026

Argus: Metric Panoramic 3D Reconstruction for Indoor Scenes

Metric feed-forward 3D reconstruction for panoramic data remains under-explored due to the lack of large-scale panoramic RGB-D training data. We present Realsee3D, a hybrid dataset of 10K indoor scenes (1K real, 9K synthetic) with 299K panoramic viewpoints and precise metric annotations, and Argus, a feed-forward network trained on it for metric panoramic 3D reconstruction. In the sparse unordered capture setting of Realsee3D, a poorly chosen coordinate anchor can cause global pose drift. Argus addresses this with a learned covisibility module that selects the geometrically optimal reference view to anchor the metric world frame. To further improve multi-task learning, we decompose the bidirectional pixel-to-world mapping into interpretable sub-steps with per-step supervision and cross-coordinate joint constraints, reinforcing geometric consistency across prediction branches. On the Realsee3D benchmark, Argus achieves state-of-the-art metric performance in camera pose estimation, depth estimation, and point cloud reconstruction. Project page: https://argus-paper.realsee.ai.

cs.CV↗

arXiv2026

Cross-Modality Feature Fusion Based on Structured State Space Duality for Multimodal Image Registration Network

In multi-modal image registration, the primary challenge lies in shared structural information extraction. Compared to Transformers, Structured State Space Duality (SSD) offers greater global structural feature extraction with higher efficiency during training and inference. Inspired by these advantages, we propose a novel algorithm for multi-modal image registration, named RegNetMamba-2. Our algorithm incorporates SSD into coarse-to-fine matching process to extract local and global structural features effectively. Firstly, SSD is applied in three different scales for multi-modal feature extraction in our network. To strengthen local representation, we pay more attention on foreground edge and structural information by feature scaling function of SSD. Secondly, for shared feature extraction of input images and multi-modal feature fusion in all scales, we propose cross-modality feature fusion model based on SSD, consisting of Cross-Modality feature Interaction (CMI) module and Multi-Scale feature Fusion (MSF) module. CMI module is designed for cross-modality feature extraction of each scale by SSD in cross form. MSF module is designed to employ a progressive upward fusion in feature-level to obtain fine features, consisting of multi-modal features in all scales. Following coarse-to-fine, the features in 1/8 scale from CMI and 1/2 scale from MSF are collected to calculate matching probability scores. Then we respectively establish matching process by correspondences of pixel-wise. Extensive experiments demonstrate that comparing with state-of-the-art deep-learning based algorithms, RegNetMamba-2 has achieved good effects in both performance and efficiency for multi-modal image registration on the following datasets: VIS-SAR (OSDataset), VIS-IR (LGHD/RoadSence) and VIS-NIR (RGB-NIR sense).

cs.CV↗

arXiv2026

Neural Radiated-Noise Fields for Unmanned Underwater Vehicle Noise Spectrum Prediction in Three-Dimensional Scenes

Radiated noise in unmanned underwater vehicles (UUVs) is an important indicator for characterizing acoustic signatures and evaluating platform performance. To address the strong dependence of traditional physics-based modeling and numerical simulation methods on target structural information and environmental boundary conditions, and their inability to achieve continuous spatial spectrum-response modeling in three-dimensional scenes, this paper proposes a neural radiated-noise field (NRNF). An NRNF represents the UUV radiated-noise spectrum as a continuous function of the three-dimensional UUV position, the three-dimensional hydrophone position, the UUV yaw angle, and the frequency, enabling query-based prediction at arbitrary spatial locations. The proposed method employs sinusoidal encoding for position and frequency, and introduces a learnable three-dimensional scene feature grid to explicitly represent environmental structure and propagation effects. A spectrum-prediction dataset is constructed from lake trials, and the proposed model is evaluated under three settings: horizontal extrapolation, depth extrapolation, and cross-run generalization. Results show that the NRNF achieves an average prediction error of 3.5 dB in the 50 to 5000 Hz band. Horizontal extrapolation is easiest, depth extrapolation is the most challenging, and cross-run generalization is of intermediate difficulty. Further ablation results demonstrate that the scene feature grid significantly improves the prediction stability and spatial generalization of the model.

eess.SP↗

arXiv2026

Non-Forgetting Knowledge Allocation with Bi-level Competition for Class-Incremental Learning

Class-Incremental Learning (CIL) with pre-trained models (PTMs) aims to sequentially adapt PTMs to new categories without forgetting old knowledge. Built upon PTMs, existing adapter-based methods mainly train models via distinct task-specific adapters, and present a uniform knowledge allocation for each adapter during inference. However, this allocation mechanism ignores the nature of task discrepancy and leads to suboptimal utilization of adapters. Also, under CIL constraint, an allocator is prone to forgetting when tasks evolve. To address these issues, we propose a Non-Forgetting Allocation with Bi-Level Competition (NoFA-BC). NoFA-BC constructs a non-forgetting allocator (NFA) by transforming the allocator training into a recursive least-squares problem and achieves an allocator equivalent to that trained with all data. Based on the NFA, a Bi-Level Competition (BLC) including an intra-task level Winner-Takes-All (WTA) mechanism and inter-task Last-Ones-Fall (LOF) elimination is proposed to provide better allocation of adapter knowledge. WTA extracts the most significant logit within a task to represent the adapter's contribution and LOF suppresses the irrelevant adapters. With BLC, participation ratio of each adapter can be tailored for each input. Moreover, a Stability Enhancement (SE) process is incorporated to further improve the performance of old tasks.

cs.CV↗

arXiv2026

StepAudio 2.5 Technical Report

Unified audio-language modeling has emerged as a prominent trend in modern speech systems, promising to bring the reasoning capabilities of large language models to auditory tasks. However, existing unified foundations often struggle to match the depth of specialized systems across automatic speech recognition (ASR), text-to-speech synthesis (TTS), and realtime spoken interaction. Bridging this gap remains an open challenge. This report presents StepAudio 2.5, a unified audio-language foundation model that matches or exceeds specialized systems across all three capabilities. Rather than treating these tasks as architecturally distinct, we operate on the premise that once text and audio share a multimodal representational space, task specialization becomes a matter of operational regimes: data construction, optimization targets, and decoding constraints. Guided by this insight, we advance the post-training paradigm from standard supervised learning to task-tailored Reinforcement Learning from Human Feedback (RLHF), using it as the primary mechanism to define complex optimization targets. We leverage this RLHF-centric alignment, alongside specialized decoding, to shape a shared backbone into three distinct operational modes. Concretely, the ASR branch advances transcription efficiency via verifiable multi-token decoding; the TTS branch achieves controllable, expressive synthesis through preference-based RLHF and context-rich supervision; and the Realtime branch realizes low-latency, persona-consistent dialogue via generative reward modeling within an RLHF framework. On standard benchmarks, StepAudio 2.5 achieves state-of-the-art results across ASR, TTS, and Realtime, demonstrating that a singular audio-language foundation can successfully internalize the distinct deployment objectives of speech understanding, generation, and live interaction.

eess.AS↗

arXiv2026

PRiMeFlow: Capturing Complex Expression Heterogeneity in Perturbation Response Modelling

Predicting the effects of perturbations in-silico on cell state can identify drivers of cell behavior at scale and accelerate drug discovery. However, modeling challenges remain due to the inherent heterogeneity of single cell gene expression and the complex, latent gene dependencies. Here, we present PRiMeFlow, an end-to-end flow matching based approach to directly model the effects of genetic and small molecule perturbations in the gene expression space. The distribution-fitting approach taken by PRiMeFlow enables it to accurately approximate the empirical distribution of single-cell gene expression, which we demonstrate through extensive benchmarking inside PerturBench. Through ablation studies, we also validate important model design choices such as operating in gene expression space and parameterizing the velocity field with a U-Net architecture. Finally, by scaling PRiMeFlow to a broad perturbation data atlas spanning multiple datasets and employing a carefully designed pretraining-finetuning strategy, we demonstrate its outstanding performance on the H1 human embryonic stem cells from the ARC Virtual Cell Challenge benchmark.

cs.LG↗

arXiv2026

NaP-Control: Navigating Diffusion Prior for Versatile and Fast Character Control

Achieving precise, versatile whole-body character control in physics-based animation remains challenging. Recent diffusion-based policies generate rich and expressive motions but typically rely on gradient-based test-time guidance to satisfy task objectives, which is slow and can reduce robustness. We introduce NaP-Control (Navigating Diffusion Prior for Versatile and Fast Character Control), abbreviated as NaP. Our method uses reinforcement learning to manipulate the latent noise of a task-agnostic diffusion policy prior, steering it toward task-specific behaviors for fast, robust control with high motion fidelity. In contrast to methods that rely solely on offline training, NaP interacts with the environment during training to correct motions and optimize task rewards, improving success rates and enabling adaptation to challenging scenarios. By directly predicting task-optimized diffusion noise, NaP eliminates iterative guidance during denoising and enables efficient inference. Experiments show that NaP attains higher success rates and faster inference while preserving natural motion across diverse tasks.

cs.GR↗

arXiv2026

Universal and Parameter-free Gradient Sliding for Composite Optimization

We propose a Parameter-Free Universal Gradient Sliding (PFUGS) algorithm for computing an approximate solution to the convex composite optimization $\min_{x\in X} \{f(x) + g(x)\}$, where $f$ has $(M_\nu,\nu)$-H\"older continuous subgradient and $g$ has $L$-Lipschitz continuous gradient. PFUGS computes an $\varepsilon$-approximate solution with $\mathcal{O}((M_\nu/\varepsilon)^{{2}/{(1+3\nu)}})$ evaluations of (sub)gradients of $f$ and $\mathcal{O}((L/\varepsilon)^{1/2})$ evaluations of gradients of $g$, without prior knowledge of problem constants. To the best of our knowledge, PFUGS is the first gradient sliding algorithm for problems involving two functions whose distinct problem constants are both unknown a priori.

math.OC↗

arXiv2026

A Note on the Gradient-Evaluation Sequence in Accelerated Gradient Methods

Nesterov's accelerated gradient descent method (AGD) is a seminal deterministic first-order method known to achieve the optimal order of iteration complexity for solving convex smooth optimization problems. Two distinct sequences of iterates are included in the description of AGD: gradient evaluations are performed at one sequence, while approximate solutions are selected from the other. The iteration complexity on minimizing objective function value has been well-studied in the literature, but such analysis is almost always performed only at the approximate solution sequence. To the best of our knowledge, for projection-based AGD that solves problems with feasible sets, it is still an open research question whether the gradient evaluation sequence (when treated as approximate solutions) could also achieve the same optimal order of iteration complexity. It is also unknown whether such results still hold in the non-Euclidean setting. Motivated by computer-aided algorithm analysis, we provide positive results that answer the open problems affirmatively. Specifically, for (possibly constrained) problem $f^*:=\min_{x\in X}f(x)$ where $f$ is convex and $L$-smooth and $X$ is closed, convex and projection friendly, we prove that the gradient-evaluation sequence $\{\underline{x}_k\}$ in AGD satisfies that $f(\underline{x}_k) - f^*\le \mathcal O(L/k^2)$.

math.OC↗

arXiv2026

Semantic-Contact Fields for Category-Level Generalizable Tactile Tool Manipulation

Generalizing tool manipulation requires both semantic planning and precise physical control. Modern generalist robot policies, such as Vision-Language-Action (VLA) models, often lack the physical grounding required for contact-rich tool manipulation. Conversely, existing contact-aware policies that leverage tactile or haptic sensing are typically instance-specific and fail to generalize across diverse tool geometries. Bridging this gap requires learning representations that are both semantically transferable and physically grounded, yet a fundamental barrier remains: diverse real-world tactile data are prohibitive to collect at scale, while direct zero-shot sim-to-real transfer is challenging due to the complex nonlinear deformation of soft tactile sensors. To address this, we propose Semantic-Contact Fields (SCFields), a unified 3D representation that fuses visual semantics with dense extrinsic contact estimates, including contact probability and force. SCFields is learned through a two-stage Sim-to-Real Contact Learning Pipeline: we first pre-train on large-scale simulation to learn geometry-aware contact priors, then fine-tune on a small set of real data pseudo-labeled via geometric heuristics and force optimization to align real tactile signals. The resulting force-aware representation serves as the dense observation input to a diffusion policy, enabling physical generalization to unseen tool instances. Experiments on scraping, crayon drawing, and peeling demonstrate robust category-level generalization, significantly outperforming vision-only and raw-tactile baselines. Project page: https://kevinskwk.github.io/SCFields/.

cs.RO↗

arXiv2026

Model-free Analysis of Scattering and Imaging Data with Escort-Weighted Shannon Entropy and Divergence Matrices

We demonstrate a model-free data analysis framework that leverages escort-weighted Shannon Entropy and several divergence matrices to detect phase transitions in scattering and imaging datasets. By establishing a connection between physical entropy and informational entropy, this approach provides a sensitive method for identifying phase transitions without an explicit physical model or order parameter. We further show that pairwise divergence matrices, including Kullback-Leibler divergence, Jeffrey Divergence, Jensen-Shannon Divergence and antisymmetric Kullback-Leibler divergence, provide more comprehensive measures of statistical changes than scalar entropy alone. Our approach successfully detects the onset of both long- and short-range order in neutron and X-ray scattering data, as well as a non-trivial phase transition in magnetic skyrmion lattices observed through Lorentz-transition electron microscopy. These results establish a framework for automated, model-free analysis of experimental data with broad applications in materials science and condensed matter physics.

cond-mat.mtrl-sci↗

arXiv2026

Noise Reduction for Pufferfish Privacy: A Practical Noise Calibration Method

This paper introduces a relaxed noise calibration method to enhance data utility while attaining pufferfish privacy. This work builds on the existing $1$-Wasserstein (Kantorovich) mechanism by alleviating the existing overly strict condition that leads to excessive noise, and proposes a practical mechanism design algorithm as a general solution. We prove that a strict noise reduction by our approach always exists compared to $1$-Wasserstein mechanism for all privacy budgets $\epsilon$ and prior beliefs, and the noise reduction (also represents improvement on data utility) gains increase significantly for low privacy budget situations--which are commonly seen in real-world deployments. We also analyze the variation and optimality of the noise reduction with different prior distributions. Moreover, all the properties of the noise reduction still exist in the worst-case $1$-Wasserstein mechanism we introduced, when the additive noise is largest. We further show that the worst-case $1$-Wasserstein mechanism is equivalent to the $\ell_1$-sensitivity method. Experimental results on three real-world datasets demonstrate $47\%$ to $87\%$ improvement in data utility.

cs.CR↗