arXiv subjects

Gang Wang

Publications and source records attributed to Gang Wang.

At least 55 records · Page 3Linked to original sources

arXiv2026

Autonomous Aerial Manipulation via Contextual Contrastive Meta Reinforcement Learning

Unmanned aerial vehicles (UAVs) are increasingly being deployed in logistics, service robotics, and other real-world applications, creating a growing demand for autonomous payload acquisition and delivery. Existing approaches typically assume pre-attached payloads or rely on specialized grippers, leaving versatile end-to-end aerial delivery largely unresolved, where different payloads induce highly variable flight dynamics, requiring a single policy to adapt online without manual calibration or explicit system identification. To this end, we study \textbf{A}utonomous \textbf{A}erial Manipulation via \textbf{Co}ntextual \textbf{Co}ntrastive Meta Reinforcement Learning (\textbf{\textit{Aco2}}), a fully autonomous aerial delivery setting in which a quadrotor equipped with a lightweight hook continuously picks up, transports, and delivers diverse handle-equipped objects between randomized locations, all without human intervention. First, we design a contextual observation encoder that infers a compact latent context from recent interaction history, enabling the policy to adapt online to payload-dependent dynamics. To further improve the quality of this context, we introduce a contrastive objective that structures the context embedding around task-relevant variations, improving generalization across diverse payloads without requiring explicit system identification. Trained entirely in simulation with extensive domain randomization, \textit{Aco2} can be directly deployed on a physical quadrotor without real-world fine-tuning.

cs.LG↗

arXiv2026

When Hard Negatives Hurt: Bridging the Generative-Discriminative Gap in Hard Negative Synthesis for Retrieval

Hard negative mining has become the dominant strategy for training retrievers, yet it faces intrinsic limitations: negatives are bounded by corpus availability, selected by retriever score rather than diagnostic value, and increasingly contaminated by false positives as the retriever improves. LLM-based synthesis offers a principled alternative, where negatives that are unconstrained, targeted, and free from false positive risk. But we show that naively incorporating generated negatives into contrastive learning often degrades retrieval performance. We identify and formalize the root cause as a generative-discriminative gap: LLM generation optimizes for fluent, plausible text, while contrastive learning demands strategic violations of relevance at the decision boundary. Our analysis reveals two compounding failure modes: discriminative-agnostic generation, where the LLM lacks an explicit model of query information needs and defaults to generic or topic-drifted text that provides no contrastive signal; and source-dependent shortcuts, where distributional artifacts enable the model to distinguish negatives by origin rather than relevance, causing gradient drift that actively corrupts optimization. To close this gap, we propose CausalNeg consisting of two main modules: (1) CoT-guided counterfactual perturbation for data construction: decomposes why a document satisfies a query into explicit information requirements, then surgically violates individual requirements to construct negatives with controlled, interpretable hardness. (2) Query-view entropy maximization during training: disperses generated negatives across the similarity spectrum, minimizing the mutual information between source identity and similarity scores to suppress shortcut exploitation. We make our code publicly available at https://github.com/mzhangzhicheng/CausalNeg.

cs.LG↗

arXiv2026

Correlation and Data-Analysis Distinctiveness of Time-Delay Interferometry Configurations

Time-Delay Interferometry (TDI) is essential for space-based gravitational wave (GW) missions, as it suppresses laser frequency noise and achieve the required sensitivity. Beyond the standard Michelson configuration, a variety of second-generation TDI schemes have been proposed, each utilizing different combinations of inter-spacecraft laser links. In this work, we conduct a comparative study of several representative TDI configurations with different time spans, and show that while their (quasi-)orthogonal channels are highly correlated, their performance in data analysis can differ among these schemes. In the low-frequency regime, the performance of different TDI configurations are nearly identical. Their distinctions emerge primarily at high frequencies, where the GW wavelength becomes comparable to the arm length. In this regime, shorter TDI time spans with minimal null frequencies facilitate more accurate waveform modeling and parameter recovery in frequency domain. In contrast, configurations with longer time spans and more null frequencies, such as the Michelson, are more susceptible to frequency aliasing and waveform modulation effects, which degrade inference accuracy. However, if signal modeling and analysis are performed in the time domain, the optimal science channels of these TDI configurations exhibit consistent performance in parameter inference. Considering the usability in both frequency and time domain, the short-span PD4L scheme, which exhibits minimal nulls and superior performance in high frequencies, emerges as a promising candidate for future space-based GW mission designs.

gr-qc↗

arXiv2026

UltraVR: A Diagnostic Ultra-Resolution Image-VQA Benchmark for Evidence-Grounded Reasoning

Vision-language models (VLMs) excel on visual question answering and multimodal reasoning benchmarks. Yet their capability on ultra-resolution images - where critical evidence is tiny, subtle, spatially distant, or distributed - remains unclear. Existing evaluations largely report final-answer accuracy, offering limited insight into whether models acquire and integrate the necessary visual evidence. We introduce UltraVR, a diagnostic benchmark for evidence-grounded visual reasoning over ultra-resolution images. UltraVR spans four high-value scenarios: CCTV surveillance, remote sensing (RS), whole-slide image (WSI) pathology, and industrial anomaly detection (AD). These domains pose complementary challenges: fine-grained object grounding in crowded CCTV scenes, long-range spatial comparison in RS, multi-scale evidence navigation in WSI, and subtle irregularity detection in repetitive industrial layouts. Beyond standard QA triples, each instance includes a structured ground-truth chain of thought with step-level questions, intermediate answers, and reasoning labels. These labels decompose reasoning into evidence grounding, local perception, quantification, evidence integration, and decision inference, enabling process-level diagnosis over black-box scoring. Using UltraVR, we evaluate frontier VLMs and show that current models remain far from reliable on ultra-resolution reasoning. Importantly, the structured annotations allow us to localize failures across the visual-to-decision pipeline: errors concentrate in evidence grounding and local perception, while downstream inference often recovers when intermediate visual facts are supplied. These findings demonstrate UltraVR as a diagnostic testbed for measuring not only whether VLMs answer correctly, but where their ultra-resolution reasoning process breaks.

cs.CV↗

arXiv2026

MAD: Mapping-Aware World Models for Agile Quadrotor Flight

Agile quadrotor flight in cluttered scenes requires more than a reactive mapping from a depth image to a control command: the vehicle must remember which regions have been observed, infer nearby occupied space, and act under partial visibility and tight latency. In this paper, we present Mapping-Aware Dreamer (MAD), a geometry-aware world model for vision-based quadrotor flight. Instead of using raw-image reconstruction as the main self-supervised objective, MAD learns recurrent latent dynamics that reconstruct robocentric occupancy and visibility grid maps together with proprioceptive states. This design forces the latent state to encode local geometry, visibility history, and ego-motion in a form that is directly relevant to collision avoidance. MAD is trained in DiffAero using a GPU-parallel map-construction module that provides high-throughput supervision for occupancy and visibility. The learned representation is used in three policy-learning modes: imagination-based MAD-Dreamer and feature-extractor variants based on PPO and SHAC. Across visual navigation and racing tasks, MAD-based agents achieve higher success rates, faster flight, and better cross-task transfer than corresponding vision-only baselines. The model also produces interpretable map predictions and accurate ego-motion estimates from depth observations. We further deploy the learned policy on a physical quadrotor with an Intel RealSense D435i and demonstrate safe indoor and outdoor flight under limited sensing, reaching 9.66 m/s in simulation and 5.05 m/s in real-world forest experiments. These results show that mapping-aware world models provide a practical middle ground between modular aerial navigation and end-to-end learning.

cs.RO↗

arXiv2026

Potential-Guided Flow Matching for Vision-Language-Action Policy Improvement

Large vision-language-action (VLA) policies are increasingly trained as conditional generative models over action chunks. Yet deployment produces mixed-quality experience-successful demonstrations, partial completions, recoverable mistakes, and failures-that is difficult to use with standard imitation. Full behavior cloning (BC) imitates failures, filtered BC discards useful sub-trajectories, and offline reinforcement learning adds a large critic. We introduce ForesightFlow, a self-guided flow-matching policy that augments each generated action chunk with a learned success-potential trajectory. The same flow proposes and scores candidate actions, enabling best-of-$K$ inference without an external critic. The key issue is that policy improvement and value calibration require different supervision: advantage weighting should emphasize high-quality actions, but applying the same weights to potential coordinates suppresses failure gradients and creates overconfident scores. We address this with decoupled advantage-weighted flow matching, applying exponentiated advantage weights only to action velocities while training potential velocities uniformly. We further derive a one-step boundary estimator for conditional flow matching, allowing advantage computation with a single stop-gradient forward pass. Across five BEHAVIOR-1K simulation tasks and five real-world bimanual tasks, ForesightFlow improves over imitation baselines, matches the strongest separate-critic baseline in simulation success, improves real-world success, and reduces training compute by $38\%$. Ablations show that decoupling prevents value hallucination, the one-step estimator preserves candidate-ranking fidelity, and self-guided sampling improves long-horizon execution.

cs.RO↗

arXiv2026

Mixture of Horizons in Action Chunking

Vision-language-action (VLA) models have shown remarkable capabilities in robotic manipulation, but their performance is sensitive to the $\textbf{action chunk length}$ used during training, termed $\textbf{horizon}$. Our empirical study reveals an inherent trade-off: longer horizons provide stronger global foresight but degrade fine-grained accuracy, while shorter ones sharpen local control yet struggle on long-term tasks, implying fixed choice of single horizons being suboptimal. To mitigate the trade-off, we propose a $\textbf{mixture of horizons (MoH)}$ strategy. MoH rearranges the action chunk into several segments with different horizons, processes them in parallel with a shared action transformer, and fuses outputs with a light linear gate. It has three appealing benefits. 1) MoH exploits long-term foresight and short-term precision jointly within a single model, improving both performance and generalizability to complex tasks. 2) MoH is plug-and-play for full-attention action modules with minimal training or inference overhead. 3) MoH enables dynamic inference with adaptive horizons, which selects stable actions through cross-horizon consensus, achieving 2.5$\times$ higher throughput than baselines while preserving superior performance. Extensive experiments over flow-based policies $π_0$, $π_{0.5}$, and one-step regression policy $π_{\text{reg}}$ demonstrate that MoH yields consistent and significant gains on both simulations and real-world tasks. Notably, under mixed-task setting, $π_{0.5}$ with MoH reaches a new state-of-the-art with 99$\%$ average success rate on LIBERO after only $30k$ training iterations. Project page: https://timsty1.github.io/moh/

cs.RO↗

arXiv2026

CCLab: Adversarial Testing of Learning- and Non-Learning-Based Congestion Controllers

Congestion controllers (CCs) are critical to network performance, and yet their robustness under adverse conditions remains insufficiently understood. While recent learning-based CCs have demonstrated strong performance in controlled environments, it is unclear how they compare to traditional CCs when controllers' input signals are corrupted or when environmental conditions become systematically challenging. In this paper, we introduce CCLab, an adversarial testing framework for systematically evaluating the robustness of both learning-based and non-learning-based CCs. CCLab includes a reinforcement learning (RL)-based adversarial agent that operates in a closed loop with the congestion control policy, generating bounded perturbations either on input signals (feature-level) or on external network conditions (environment-level), while preserving realism through explicit constraints. Using this framework, we compare learning-based CCs with non-learning-based CCs under both feature-level and environment-level adversarial conditions. While both types of CCs suffer from performance degradation under adversarial testing, we find that learning-based CCs, in general, are more robust than traditional human-designed algorithms. Finally, we show that our adversarial traces can be used to train more robust CCs that outperform existing learning-based CCs under both challenging and normal conditions.

cs.CR↗

arXiv2026

Physics-Informed Generative Solver: Bridging Data-Driven Priors and Conservation Laws for Stable Spatiotemporal Field Reconstruction

Reconstructing continuous physical fields from sparse measurements is a central inverse problem, but data-driven generative models can produce states that violate governing dynamics. We introduce a physics-informed generative solver that separates stable prior learning from inference-time enforcement of conservation laws. Martingale-Regularized Score Matching regularizes score pretraining with a Score Fokker-Planck constraint, yielding a dynamically stable prior. Physics-Informed Implicit Score Sampling then guides denoising trajectories by gradients of physical residuals, projecting samples toward admissible manifolds without retraining. In acoustics, the method co-generates pressure and particle velocity from sparse sensors, enabling dense virtual arrays that suppress spatial aliasing. The same framework generalizes to real-world ERA5 meteorological fields under extreme sparsity. Together, this work establishes a rigorous and generalizable paradigm for solving high-dimensional inverse problems, bridging the gap between generative artificial intelligence and first-principles science.

cs.LG↗

arXiv2026

Entangled Telecom Photon Generation using Twisted Van der Waals Crystals

Nanoscale quantum light sources are essential building blocks for integrated quantum photonic systems. Here, we report a wavelength-scale entangled-photon source based on van der Waals-engineered NbOBr$_2$, and benchmark its performance for telecom-wavelength quantum light generation. By exploiting the material's second-order nonlinearity, we generate quantum-correlated photon pairs via spontaneous parametric down-conversion. We then use a 90$^{\circ}$ twisted stacking to induce quantum interference in photon-pair generation, yielding polarization-entangled photons. This approach enables tunability of the quantum optical state via control of the excitation laser polarization. We experimentally obtain entanglement fidelities exceeding 95% for Bell states, along with a high coincidence-to-accidental ratio of $\sim$335, and a brightness approximately one order of magnitude higher than recently reported telecom sources based on transition metal dichalcogenide (TMD) 2D materials. These results establish twisted van der Waals engineering as a powerful platform for highly tunable, high-brightness quantum light sources at telecom wavelengths.

physics.optics↗

arXiv2026

CoT-Guard: Small Models for Strong Monitoring

Monitoring the chain-of-thought (CoT) of reasoning models is a promising approach for detecting covert misbehavior (i.e., hidden objectives) in code generation tasks. While large models (GPT-5, Gemini-3-Flash) can serve as effective CoT monitors, they are expensive to deploy due to the lengthy reasoning traces and high API cost, emphasizing the need for smaller, cheaper alternatives. Nevertheless, we find that current small models (4B--8B) struggle to detect hidden objectives despite access to the CoT, frequently misattributing them as part of the user query. To address this, we propose a post-training pipeline combining supervised fine-tuning (SFT) and reinforcement learning (RL), where SFT narrows the gap for in-domain tasks by distilling detection behavior from stronger monitors, and RL on hard and subtly crafted hidden objectives helps the model generalize to out-of-domain monitoring tasks. To validate this generalization, we evaluate under a realistic threat model motivated by practical supply-chain attacks, where the adversary is a third-party LLM router injecting hidden objectives into code-generation requests through either prompt manipulation or code manipulation attacks. To push beyond objectives that large monitors already saturate, we also introduce four new challenging tasks even for strong monitors. Finally, we introduce CoT-Guard, a 4B-parameter monitor that demonstrates superior generalization performance under both prompt and code manipulation attacks, achieving a G-mean^2 (i.e., TNR x TPR) of 75% and outperforming GPT-5.4 (56%), GPT-5-mini (41%), and Qwen3-32B (54%), while closing the gap to Gemini-3-Flash (83%). These results demonstrate that CoT-Guard provides a practical and cost-effective user-side defense, substantially improving hidden-objective detection while avoiding the deployment cost of large monitors.

cs.CR↗

arXiv2026

Towards Autonomous Business Intelligence via Data-to-Insight Discovery Agent

Transforming fragmented enterprise data into actionable insights remains a significant challenge for LLMs, constrained by complex database schemas, limitations in dynamic SQL generation, and the need for deep multi-dimensional analysis.In this paper, we propose AIDA(Autonomous Insight Discovery Agent), the first end-to-end framework designed for autonomous exploration in complex business environments. We establish a highly flexible instant retail environment encompassing 200+ metrics and 100+ dimensions, and integrates a proprietary Domain-Specific Language (DSL) that bridges semantic reasoning with precise SQL execution. Our reinforcement learning system subsequently formulates business analysis as a Pareto Principle-guided cumulative reasoning process. Experimental results demonstrate that AIDA significantly outperforms workflow-based agents, and extensive evaluations further reveal that AIDA achieves superior environmental perception and more in-depth analysis from diverse perspectives. Our work ultimately establishes the transformative potential of autonomous intelligence for industrial-scale business intelligence systems.

cs.AI↗

arXiv2026

FollowTable: A Benchmark for Instruction-Following Table Retrieval

Table Retrieval (TR) has traditionally been formulated as an ad-hoc retrieval problem, where relevance is primarily determined by topical semantic similarity. With the growing adoption of LLM-based agentic systems, access to structured data is increasingly instruction-driven, where relevance is conditional on explicit content and schema constraints rather than topical similarity alone. We therefore formalize Instruction-Following Table Retrieval (IFTR), a new task that requires models to jointly satisfy topical relevance and fine-grained instruction constraints. We identify two core challenges in IFTR: (i) sensitivity to content scope, such as inclusion and exclusion constraints, and (ii) awareness of schema-grounded requirements, including column semantics and representation granularity--capabilities largely absent in existing retrievers. To support systematic evaluation, we introduce FollowTable, the first large-scale benchmark for IFTR, constructed via a taxonomy-driven annotation pipeline. We further propose a new metric, termed the Instruction Responsiveness Score, to evaluate whether retrieval rankings consistently adapt to user instructions relative to a topic-only baseline. Our results indicate that existing retrieval models struggle to follow fine-grained instructions over tabular data. In particular, they exhibit systematic biases toward surface-level semantic cues and remain limited in handling schema-grounded constraints, highlighting substantial room for future improvements.

cs.IR↗

arXiv2026

Echo-α: Large Agentic Multimodal Reasoning Model for Ultrasound Interpretation

Ultrasound interpretation requires both precise lesion localization and holistic clinical reasoning, yet existing methods typically excel at only one of these capabilities: specialized detectors offer strong localization but limited reasoning, whereas multimodal large language models (MLLMs) provide flexible reasoning but weak grounding in specialized medical domains. We present Echo-α, an agentic multimodal reasoning model for ultrasound interpretation that unifies these strengths within an invoke-and-reason framework. Echo-α is trained to coordinate organ-specific detector outputs, integrate them with global visual context, and convert the resulting evidence into grounded diagnostic decisions beyond detector-only inference. This behavior is established through a nine-task supervised curriculum and then refined by sequential reinforcement learning under different reward trade-offs, yielding Echo-α-Grounding for lesion anchoring and Echo-α-Diagnosis for final diagnosis. On multi-center renal and breast ultrasound benchmarks, Echo-α outperforms competitive baselines on both grounding and diagnosis. In particular, on cross-center test sets, Echo-α-Grounding attains 56.73%/43.78% F1@0.5 and Echo- α-Diagnosis reaches 74.90%/49.20% overall accuracy on renal/breast ultrasound. These results suggest that agentic multimodal reasoning can turn specialized detectors into verifiable clinical evidence, offering a practical route toward ultrasound AI systems that are more accurate, interpretable, and transferable. The repository is at https://github.com/MiliLab/Echo-Alpha.

cs.CV↗

arXiv2026

Probing Jet-Medium Interactions in Heavy-Ion Collisions Using Energy-Energy Correlators

Energy-energy correlators (EECs) provide a sensitive probe of both perturbative and nonperturbative dynamics in relativistic heavy-ion collisions. Jet-medium interactions enhance particle multiplicity within the jet cone, which must be properly accounted for when extracting the EEC of jet shower hadrons in experiments. To address this issue, we develop an augmentation method that exploits momentum conservation between the near-side and away-side regions, using $γ$-jet events with 0-10\% centrality in Pb+Pb collisions at $\sqrt{s_{NN}} = 5.02$ TeV simulated with the CoLBT-hydro model. This approach yields an experimentally reconstructed EEC that shows improved agreement with the EEC of hadrons originating primarily from jet parton splittings. Comparing EECs of jets from Pb+Pb and p+p collisions with different matching conditions can be sensitive to jet medium interaction dynamics, and provide a novel means to test the scenario of jet energy loss in the QGP, followed by fragmentations outside the QGP.

nucl-th↗

arXiv2026

CPI-C: Cool Planet Imaging Coronagraph on Chinese Space Station Survey Telescope

Cool Planet Imaging Coronagraph (CPI-C) on Chinese Space Station Survey Telescope (CSST) is proposed to direct image the cool planets around nearby solar-type stars (within 40 pc). The core scientific objective of CPI-C is to conduct high-contrast directly imaging surveys of exoplanets ranging in size from Neptune-like to Jupiter-like, located at separations of 0.5 to 5 AU from their host stars, and to perform systematic spectroscopic analysis of the detected planets through high-precision multi-band photometry. CPI-C employs a step-transmission apodization technique to suppress the diffraction noises from the telescope pupil and a precise phase correction technique to eliminate the speckle noises due to imperfections of the optical surfaces. The contrast requirement is better than $10^{-8}$ at an inner working angle (IWA) of $3-4λ/D$, in the visible wavelength from 600 nm to 900 nm. CPI-C will be the first space-based instrument capable of directly imaging the reflection light from the cool exoplanets in the visible wavelength enabling the measurement of key physical parameters such as the effective temperature, surface gravity, radius, mass, and other key parameters. The potential observation results will significantly contribute to further understand the formation and evolution mechanisms of planets, which will also lay a solid foundation for future confirmation of the Earth-twins in the next generation space flagship missions.

astro-ph.EP↗

arXiv2026

Wan-Image: Pushing the Boundaries of Generative Visual Intelligence

We present Wan-Image, a unified visual generation system explicitly engineered to paradigm-shift image generation models from casual synthesizers into professional-grade productivity tools. While contemporary diffusion models excel at aesthetic generation, they frequently encounter critical bottlenecks in rigorous design workflows that demand absolute controllability, complex typography rendering, and strict identity preservation. To address these challenges, Wan-Image features a natively unified multi-modal architecture by synergizing the cognitive capabilities of large language models with the high-fidelity pixel synthesis of diffusion transformers, which seamlessly translates highly nuanced user intents into precise visual outputs. It is fundamentally powered by large-scale multi-modal data scaling, a systematic fine-grained annotation engine, and curated reinforcement learning data to surpass basic instruction following and unlock expert-level professional capabilities. These include ultra-long complex text rendering, hyper-diverse portrait generation, palette-guided generation, multi-subject identity preservation, coherent sequential visual generation, precise multi-modal interactive editing, native alpha-channel generation, and high-efficiency 4K synthesis. Across diverse human evaluations, Wan-Image exceeds Seedream 5.0 Lite and GPT Image 1.5 in overall performance, reaching parity with Nano Banana Pro in challenging tasks. Ultimately, Wan-Image revolutionizes visual content creation across e-commerce, entertainment, education, and personal productivity, redefining the boundaries of professional visual synthesis.

cs.CV↗

arXiv2026

Spin Fluctuations in the Rare-Earth Doped Bilayer Nickelates

Spin fluctuations have been generally believed as the pairing glue of high-$T_c$ superconductivity. Recent inelastic neutron scattering (INS) studies have revealed a weak flat spin-fluctuation signal around 45 meV in the bilayer nickelate La$_3$Ni$_2$O$_{7-δ}$, suggesting strong interlayer and weak intralayer magnetic couplings ($SJ_{\perp}\approx$ 60 meV, $SJ_{\parallel}\leq$ 3.5 meV) in contrast to cuprate and pnictide superconductors. Here, we report further INS studies on the Pr and Nd doped La$_3$Ni$_2$O$_{7-δ}$ powder samples at ambient pressure. Besides the crystalline electric field excitations at low energies, we have found that the 45 meV flat mode splits into two modes in doped compounds, along with another weak mode at about 60 meV, where the spin fluctuations in La$_2$NdNi$_2$O$_{7-δ}$ are stronger than La$_3$Ni$_2$O$_{7-δ}$ and La$_2$PrNi$_2$O$_{7-δ}$. Our results are consistent with an enhanced interlayer coupling $SJ_{\perp}$ within the stripe-type Heisenberg model framework, where the estimated $SJ_{\perp}$ value is in the range of about 69 to 73 meV for the rare-earth doped bilayer nickelates.

cond-mat.supr-con↗