arXiv subjects

Qi Wang

Publications and source records attributed to Qi Wang.

At least 37 records · Page 2Linked to original sources

arXiv2026

A Stochastic LineseArch Method (SLAM) for Minimizing Expectation Residuals

Most existing rate and complexity guarantees for stochastic gradient methods in $L$-smooth settings mandates that such sequences be non-adaptive, non-increasing, and upper bounded by $\tfrac{a}{L}$ for $a > 0$. This requires knowledge of $L$ and may preclude larger steps. Motivated by these shortcomings, we present an Armijo-enabled stochastic linesearch framework with common random number (CRN)-enabled stochastic zeroth- and first-order oracles. The resulting steplength sequence is non-monotone and requires neither knowledge of $L$ nor any other problem parameters. We then prove that the expected stationarity residual diminishes at a rate of $\mathcal{O}(1/\sqrt{K})$, where $K$ denotes the iteration budget. Furthermore, the resulting iteration and sample complexities for computing an $ε$-stationary point are $\mathcal{O}(ε^{-2})$ and $\mathcal{O}\left(ε^{-4}\right)$. The proposed method allows for a simple nonsmooth convex component in the objective, addressed through proximal gradient updates. Analogous guarantees are provided in the Polyak-Lojasiewicz (PL) setting and convex regimes. Preliminary numerical experiments are seen to be promising.

math.OC↗

arXiv2026

SkillPoison: Progressive Skill Poisoning via Successful Experiences

Self-improving LLM agents increasingly distill successful experiences into persistent, reusable skills. Existing skill attack methods corrupt this learning pipeline by injecting malicious triggers, behaviors, or false facts into individual experiences or extracted skills. However, such attacks are easily detected, and the injected malicious behaviors often fail to accumulate as persistent skills. In this paper, we show that skill poisoning can arise even from verified successful experiences, without making any individual trajectory malicious. Based on this insight, we propose SkillPoison, a novel framework that progressively poisons skill via successful experiences. SkillPoison first constructs a set of successful experiences that reinforce a target behavior, and then removes the contextual conditions that constrain when the behavior applies. Rather than injecting malicious content, SkillPoison shapes how the skill extractor generalizes, allowing useful behavior to support task success while inducing harmful behavior when they are misapplied. Extensive experiments on three benchmarks show that SkillPoison achieves 95.71% attack success rates, while all injected experiences remain task-correct and pass verification and lexical inspection. Our code, data and implementation details are available for the community at https://github.com/DEEP-JLU/SkillPoison.

cs.CR↗

arXiv2026

Imagine to Act: High-Fidelity Data Synthesis via Image Editing World Model for Scalable GUI Agent Training

Graphical User Interface (GUI) agents have emerged as a promising paradigm for automating complex digital workflows across diverse applications. However, training highly capable and generalizable agents fundamentally relies on massive, high-fidelity visual-action trajectories, which are notoriously difficult to acquire. While human demonstrations are unscalable, existing GUI world models rely on text descriptions or HTML rendering, discarding crucial pixel-level visual details like icons and layout styles. To address this issue, we introduce Infinite-Dreamer, a simulation-free data synthesis method powered by a pixel-level Image Editing World Model. By conceptualizing GUI transitions as image editing tasks, we leverage Vision-Language Models (VLMs) to describe action-induced UI changes as structured delta-text. We then fine-tune an image editing backbone to controllably synthesize realistic screenshot transitions. We utilize this model to generate both single-frame visual robustness data and multi-step imaginary trajectories. To validate the effectiveness of our approach, we fine-tune the Qwen3-VL baseline solely on the synthesized data to obtain Infinite-Actor, and evaluate it on AndroidWorld, MobileWorld, and AndroidControl-Curated benchmarks. Infinite-Actor consistently outperforms the Qwen3-VL baselines across scales: Infinite-Actor-8B improves AndroidWorld Pass@1 by +4.45 and nearly doubles the MobileWorld Pass@3 success rate, while Infinite-Actor-2B improves Pass@1 by +9.05. Code is available at https://github.com/swaydy-n/Infinite-Dreamer.

cs.CV↗

arXiv2026

Casual Flash Lighting for Gaussian Splat Inverse Rendering

Recovering geometry, materials, and lighting from photographs is highly ambiguous when only static illumination is available. Active-lighting setups reduce the ambiguity but require dark rooms or specialized hardware. Instead, we synergize both static and flash lighting from casual indoor capture, with the flash on or off, each from independent viewpoints. The flash residual constrains albedo and the BRDF, while static lighting captures grazing-angle specular highlights that flash misses. With a 2DGS reconstruction framing, our key contribution is a GS-anchored diffuse field: a hash-encoded MLP is queried at the rasterized 2DGS depth. As it depends only on world position, it is view consistent in 3D and allows the flash residual to drive material decomposition instead of being absorbed by alpha-blending drift across views. At the same time, we render static lighting with deferred shading such that it can also supervise material decomposition. On five synthetic and three real indoor scenes, our method outperforms six recent baselines on diffuse color, albedo and roughness material parameters, and in relighting where PSNR improves by 4.17 dB over the next-best baseline.

cs.CV↗

arXiv2026

Mapping E-textiles Design Pain Points and Generative AI Opportunities: Insights from Workshops in Shanghai and Winchester

E-textile design involves complex decisions across materials, sensor and actuator structures, fabrication, garment integration, and data processing. It typically requires iterative prototyping and testing, which are time- and labour-intensive, while few practitioners possess cross-disciplinary expertise across all relevant domains. To identify current bottlenecks and explore how Generative AI (GenAI) might support the design process, we conducted two half-day co-design workshops, one in Shanghai and one in Winchester, with practitioners from materials science, electronics, garment design, human-computer interaction, and manufacturing. Twenty practitioners participated in the Shanghai workshop; ten of them had prior experience in e-textiles and form the contributing sample analysed here. A further ten practitioners participated in the Winchester workshop. Participants mapped their own design pipelines, annotated bottlenecks, and proposed where GenAI could provide support. Rather than presenting a ranked list of opportunities, we report a process map that indexes each proposed GenAI role to the pipeline stage at which practitioners located it, together with the conditions on which they stated its usefulness would depend. Across both sites, practitioners consistently identified domain-specific operational barriers, including data scarcity, the disconnect between prototyping and manufacturing, and trade-offs in material-hardware integration. They also emphasized that the primary barrier to GenAI-driven e-textile design is not general model capability, but the lack of standardized, machine-readable representations of e-textile designs. Based on these findings, we identify four classes of domain-tailored AI tools that could support future e-textile design processes.

cs.HC↗

arXiv2026

VoCa: Designing Speech-Canvas Interaction for Voice-Based Conversational Agents

People write and sketch while speaking to explain, organize, and develop content together. Inspired by these practices, we investigate how voice agents can use a canvas alongside speech in multi-turn conversations with users. We conducted a two-part formative study: an observational study of how pairs coordinated speech and boardwork, followed by a design workshop that informed a design space for speech-canvas interaction with voice agents. Building on these insights, we developed VoCa, a voice agent that coordinates speech with visual object creation, annotation, and attention guidance. A five-day deployment with 18 participants examined usability, experiences of speech-canvas interaction, patterns of use, and desired improvements. Participants' experiences highlighted opportunities for speech-canvas interaction in learning, work, and daily life, alongside challenges in coordinating what agents say and show in ways users can follow and influence. These findings inform how voice agents can use a canvas alongside speech in conversation.

cs.HC↗

arXiv2026

Latent-space variational data assimilation in two-dimensional turbulence

Starting from limited measurements of a turbulent flow, data assimilation (DA) attempts to estimate all the spatio-temporal scales of motion. Success is dependent on whether the system is observable from the measurements, or how much of the initial turbulent field is encoded in the available measurements. Adjoint-variational DA minimises the discrepancy between the true and estimated measurements by optimising the initial velocity or vorticity field (the `state space'). Here we propose to instead optimise in a lower-dimensional latent space which is learned by implicit rank minimising autoencoders. Assimilating in latent space, rather than state space, redefines the observability of the measurements and identifies the physically meaningful perturbation directions which matter most for accurate prediction of the flow evolution. When observing coarse-grained measurements of two-dimensional Kolmogorov flow at moderate Reynolds numbers, the proposed latent-space DA approach estimates the full turbulent state with a relative error improvement of two orders of magnitude over the standard state-space DA approach. The small scales of the estimated turbulent field are predicted more faithfully with latent-space DA, greatly reducing erroneous small-scale velocities typically introduced by state-space DA. Furthermore, latent-space DA is demonstrated to be robust to noisy measurements at the range of Reynolds numbers considered. These findings demonstrate that the observability of the system from available data can be greatly improved when turbulent measurements are assimilated in the right space, or coordinates.

physics.flu-dyn↗

arXiv2026

Exchange-mediated directional coupling of short-wavelength spin waves

Miniaturizing magnonic circuits and increasing their processing speed require spin waves with progressively shorter wavelengths. However, conventional directional couplers rely on the dynamic dipolar interaction between spatially separated waveguides, which rapidly weakens in the exchange-dominated regime. As a result, their coupling length increases as the spin-wave wavelength decreases, limiting further device miniaturization. Here, we introduce an exchange-mediated directional coupler by replacing the nonmagnetic gap between two waveguides with a magnetically modified YIG spacer. The spacer provides a continuous exchange pathway and preserves the splitting between the symmetric and antisymmetric modes at large wavenumbers. We develop an analytical model to describe this mechanism and verify it using micromagnetic simulations. Based on this concept, we design a straight coupler without curved access waveguides, with a footprint of only 400 nm * 30 nm. For a spin-wave wavelength of approximately 100 nm, the device transfers approximately 98% of the normalized output power to the target waveguide. Its performance remains nearly unchanged even when the damping of the spacer is strongly increased. These results overcome the short-wavelength limitation of dipolar couplers and provide a route toward highly integrated magnonic circuits for information processing.

physics.app-ph↗

arXiv2026

StreamRig: Exploiting Intra-Rig Geometry for Streaming Multi-Camera Odometry

Mobile robots and vehicles carry synchronized multi-camera rigs, yet many streaming 3D foundation models are designed for monocular input, leaving efficient use of rig geometry a challenge. We present StreamRig, a freeze-and-stream framework that builds causal streaming odometry for calibrated rigs on a frozen multi-view 3D foundation model. The frozen front-end jointly perceives the synchronized views using rig calibration. A Rig-Resampler compresses their features, a CausalBridge applies causal attention with a key-value cache, and a lightweight head regresses rig poses. A periodic re-anchoring protocol supports stable pose estimation over long sequences. Only these modules are trained, 74.6M parameters in total, with relative poses as the sole supervision. Our two-stage training strategy combines group relocalization pretraining with causal rig training to transfer the geometric priors of the frozen front-end and the alignment ability of the pretrained modules to streaming odometry. We evaluate on NCLT, TartanGround, KITTI-360, and our self-collected humanoid-robot dataset ZJH, where training uses only simulation and real-world evaluation is zero-shot. Across all four datasets, StreamRig achieves lower translation and rotation drift than the evaluated non-oracle monocular streaming and rig-aware offline models, while maintaining low inference cost. Ablations and controlled camera-count experiments identify the sources of these gains. We further examine how longer training windows affect inference over longer horizons. Code has been released at https://github.com/WeiYuFei0217/StreamRig.

cs.CV↗

arXiv2026

Semantic Map Sharing and Capability-Aware Coverage Planning for AI-Native 6G Robotic Coordination

Search and Rescue (SAR) operations increasingly deploy heterogeneous teams of aerial and ground robots. However, conventional coverage methods typically do not translate perceived terrain into platform-specific reachability, while continuous image exchange imposes a high communication cost. We propose an edge-centric, semantic-aware coverage planning framework that integrates aerial terrain perception, robot-specific traversability reasoning, and payload-efficient semantic state sharing. Aerial observations are converted into compact semantic grid maps, enabling reachability-constrained area decomposition and capability-aware coverage paths that assign only regions admitted by each robot's capability profile. The resulting perception-sharing-planning loop feeds semantic corrections into traversability reasoning and replanning, forming an application-level mechanism motivated by AI-enabled goal-oriented communication envisioned for AI-native 6G networks. For the high-update case, transmitting semantic corrections reduces the application payload by a factor of approximately $82$ relative to periodic full-map sharing. Across matched benchmark scenarios, the proposed method achieved $91.5\%$ coverage with no capability-infeasible allocations, compared with $78.8\%$ coverage and a $21.5\%$ capability-infeasible allocation rate for LS-MCPP. Semantic corrections update the shared planning state without requiring repeated transmission of the complete map.

cs.RO↗

arXiv2026

A Bounded Degree SOS Plus SONC Hierarchy for Polynomial Optimization

We propose a bounded degree SOS+SONC hierarchy for constrained polynomial optimization, termed B-SOS+SONC. Starting from Lasserre's bounded-degree SOS framework, we enlarge the certificate cone from SOS to the recently introduced SOS+SONC cone, thereby combining the algebraic strength of semidefinite relaxations with the sparse structure captured by circuit polynomials. We show that, for each fixed certificate degree, the resulting hierarchy is complete, that is, its optimal values are monotone and converge to the global optimum. Moreover, we derive an explicit SDP-REP reformulation, so that each relaxation can be solved within a tractable convex optimization framework over semidefinite and relative entropy cones. Beyond the optimization hierarchy itself, we investigate structural properties of the SONC cone and introduce the notions of first-order and second-order SONC-convexity. This leads to a new sufficient condition for first-level exactness of the B-SOS+SONC hierarchy. Numerical experiments illustrate that the proposed hierarchy often yields tighter lower bounds than the B-SOS relaxation while remaining tractable.

math.OC↗

arXiv2026

Learning Transferable Self-Supervised Priors for Super-Resolution Reconstruction in Structured Illumination Microscopy

Structured illumination microscopy (SIM) extends the optical passband, and reconstruction of detail beyond it depends on prior knowledge. Hand-designed regularizers depend on how well their structural assumptions match the specimen; learned priors can be sensitive to changes in imaging conditions and specimen structure. We introduce SIMAdapter, which pretrains a network that predicts the emitter and the point-spread function (PSF) by self-supervision on 23,237 raw SIM stacks from BioSR, BioTISR, and simulations spanning different PSFs and specimen structures, then adapts it to a single unlabeled target stack. Adaptation refines the network against a differentiable image-formation model, with the light pattern calibrated from that stack. Both stages take their supervision from the raw measurements and need no paired high-resolution reference. On two held-out synthetic domains, SIMAdapter reaches a mean emitter normalized root-mean-square error of 0.156, compared with 0.403 for Sparse-SIM. The same adaptation started from a network pretrained on BioSR alone is less accurate in both domains. In three experimental case studies, adaptation reduces flanking artifacts and yields more distinct profiles across filament pairs, mitochondrial boundaries, and calibration lines. A single pretrained network can thus be reused across SIM measurements, with each reconstruction refined against its own raw data.

physics.optics↗

arXiv2026

LimiX-2: A Contextual Mechanism Network Towards General Structured-Data Intelligence

We introduce LimiX-2, a new model in the LimiX family, developed through model and data scaling guided by our previously established scaling laws. LimiX-2 adopts the Contextual Mechanism Networks (CMNs) paradigm and is pretrained with Context-Conditional Masked Modeling (CCMM). CMNs shifts the organizing principle of in-context learning from target-centric prediction to mechanism-oriented joint modeling. Rather than centering the network on the $p(y \mid x, D_{\mathrm{context}})$ objective of conventional tabular PFNs, it is designed around learning $p(x, y \mid D_{\mathrm{context}})$, a context-dependent representation of the joint structure underlying data generation. Pretraining uses synthetic datasets generated by structural causal models (SCMs) spanning diverse graph structures, functional mechanisms, and observation processes. Evaluations on TabArena, TALENT, and BCCO show that LimiX-2 outperforms current dataset-specific models and tabular foundation models. Beyond predictive performance, the CMN paradigm also promotes causal awareness in LimiX-2: its feature attention encodes direct causal relationships, enabling accurate causal skeleton recovery.

cs.AI↗

arXiv2026

MoCapAnything V2: End-to-End Motion Capture for Arbitrary Skeletons

Recent methods for arbitrary-skeleton motion capture from monocular video follow a factorized pipeline, where a Video-to-Pose network predicts joint positions and an analytical inverse-kinematics (IK) stage recovers joint rotations. While effective, this design is inherently limited, since joint positions do not fully determine rotations and leave degrees of freedom such as bone-axis twist ambiguous, and the non-differentiable IK stage prevents the system from adapting to noisy predictions or optimizing for the final animation objective. In this work, we present the first fully end-to-end framework in which both Video-to-Pose and Pose-to-Rotation are learnable and jointly optimized. We observe that the ambiguity in pose-to-rotation mapping arises from missing coordinate system information: the same joint positions can correspond to different rotations under different rest poses and local axis conventions. To resolve this, we introduce a reference pose-rotation pair from the target asset, which, together with the rest pose, not only anchors the mapping but also defines the underlying rotation coordinate system. This formulation turns rotation prediction into a well-constrained conditional problem and enables effective learning. In addition, our model predicts joint positions directly from video without relying on mesh intermediates, improving both robustness and efficiency. Both stages share a skeleton-aware Global-Local Graph-guided Multi-Head Attention (GL-GMHA) module for joint-level local reasoning and global coordination. Experiments on Truebones Zoo and Objaverse show that our method reduces rotation error from ~17 degrees to ~10 degrees, and to 6.54 degrees on unseen skeletons, while achieving ~20x faster inference than mesh-based pipelines. Project page: https://animotionlab.github.io/MoCapAnythingV2/

cs.CV↗

arXiv2026

Policy and World Modeling Co-Training for Language Agents

Reinforcement learning (RL) improves large language model (LLM) agents by teaching them which actions lead to high rewards, but provides little supervision on what those actions do to the environment. World modeling (WM) can fill this gap, yet existing approaches often require separate simulators, extra training stages, or additional inference-time computation. We observe that on-policy RL rollouts already contain the needed signal: each transition pairs an action with its resulting next observation. Based on this observation, we propose PaW, a Policy and World modeling co-training framework that adds auxiliary WM supervision to the same policy during RL, without changing the inference paradigm. To make auxiliary WM supervision informative and stable, PaW introduces three components: action-entropy-based WM data selection, noise-tolerant WM loss, and reward-adaptive loss balancing. Experiments on three agentic task benchmarks show consistent improvements over strong RL baselines across models and RL algorithms. These results suggest that standard RL rollouts are a practical source of WM supervision for language-agent training.

cs.LG↗

arXiv2026

Observation of topological surface phonons in diamond with nonlinear optics

Topological quantum states in electronic systems have profoundly transformed the understanding of phases of matter. Recent theories predict novel vibrational topological quantum states, i.e., topological phonons in various solids. However, this paradigm is yet to be established due to the lack of convincing experimental verification of topological surface phonons---a hallmark signature of topological phonon states. Here, we report the discovery of topological surface phonons in diamond using sum-frequency spectroscopy---a nonlinear optical spectroscopy capable of probing surface phonons with high sensitivity. Diamond, known for exceptional hardness and thermal transport, is unveiled as a phonon topological semimetal hosting topological nodal-lines and nexus triple points in the bulk. With consistent theory and experiments, we uncover the resultant topological surface phonons on diamond (111) and (100) surfaces. Moreover, by chemically modifying these surfaces, we reveal the disorder effect on topological surface phonons. These findings pave the way for bridging two fundamental domains: quantum topology and lattice dynamics, besides having important implications on diamond-based devices and functional interfaces.

cond-mat.mtrl-sci↗

arXiv2026

HaWMPO: Hallucination-Aware World Model-based Policy Optimization for Generalist Robot Policy

Generalist robot policies have demonstrated strong generalization across robotic manipulation tasks, yet their success rates remain limited in com- plex long-horizon scenarios. Recent methods improve Visual-Language-Action (VLA) policies through online reinforcement learning on real robots, but such training relies on costly physical interactions, suffers from low sample efficiency, and may introduce hardware and safety risks. World models offer a promising alternative by enabling policy optimization with imagined rollouts. However, long-horizon rollouts generated by world models often suffer from prediction hal- lucinations, producing biased state transitions that can mislead policy learning. To address this issue, we propose Hallucination-aware World Model-based Pol- icy Optimization (HaWMPO), a closed-loop reinforcement learning pipeline for VLA policy post-training with world models. Specifically, HaWMPO introduces an action-conditioned hallucination-aware model to estimate the reliability of gen- erated image sequences, and incorporates hallucination scores into group relative policy optimization through a Reward-Soft mechanism, suppressing unreliable ac- tion chunks during training. On the LIBERO benchmark, HaWMPO achieves the best average success rate, with gains of 15.0% over the base model and 2.8% over the strongest baseline; real-world experiments on a G1 robot further validate its effectiveness, raising the average success rate on two manipulation tasks from 67.5% to 80.0%.

cs.RO↗

arXiv2026

Decoupled Self-Forcing Distillation for Streaming Talking Head Generation

Streaming talking-head generation produces each frame as its driving audio arrives, yet fidelity and efficiency have so far pulled in opposite directions: end-to-end methods condition a video diffusion model on audio directly and achieve high quality but only at large scale, while cheaper two-stage methods generate an intermediate motion representation and trail in fidelity. We argue the cost of the former lies in the target of fusion: the video latent is dominated by identity, appearance and background, none of which audio bears on, so coupling audio to every pixel blurs detail and wastes capacity. We instead fuse conditions in a low-dimensional identity-disentangled motion space, routing audio and motion captions by their temporal granularity, and generate motion latents with a small causal autoregressive transformer that a pretrained diffusion renderer turns into video. Conditions thus control video transitively, and high fidelity no longer requires a large backbone. Streaming this decomposition needs both models to be causal, and the exposure-bias problem could be solved by self-forcing given a bidirectional teacher. But there is no such teacher in motion space. Our decoupled self-forcing distillation resolves both models under one frozen teacher: conditioned on motion, it distills the renderer into a block-causal student; unconditionally, it scores rendered rollouts against real videos, supervising motion by the video it produces. This lifts the fidelity ceiling from the motion generator onto the stronger renderer. The two models run as parallel causal streams, reaching 15.4 FPS at 1.3 s latency with no quality degradation.

cs.CV↗