Search arXiv⌕ Search

arXiv subjects

Yang Wang

Publications and source records attributed to Yang Wang.

At least 37 records · Page 2Linked to original sources
arXiv2026

Bringing Generative Learning to Representation Learning: Self-Supervised Transfer Learning as Distribution Matching

Most self-supervised learning objectives defend against collapse but leave the target representation law unspecified. We formulate representation learning as Distribution Matching (DM), learning an augmentation-invariant encoder whose push-forward law matches an explicit reference law. The reference law specifies what the learned representation distribution should look like, while the matching mechanism determines how this target is matched. Within DM, generative learning and representation learning can be viewed as directionally inverse distributional processes: generative learning maps a tractable reference to data, whereas representation learning maps data to a designed reference law. For the Mallows-distance instantiation studied here, we connect the population objective to class-centre separation and classification error and derive a finite-sample bound on downstream classification error. Simulations and image benchmarks show manifold rectification, fine-grained structure and transferability.

stat.ML↗
arXiv2026

Reading Between the Frames of Multimodal Drivelology

Social media videos often communicate meanings that go beyond their visible actions, captions, or speech. A mundane clip may become humorous, ironic, or satire only through the interaction of multimodal cues and cultural context, making such content a difficult test case for video-language models. In this paper, we introduce DrivelHub+, a benchmark for evaluating whether models can infer the implicit, non-linear, and rhetorically layered meanings of social media videos that appear nonsensical on the surface but convey deliberate pragmatic meanings. DrivelHub+ consists of 1,000 videos collected from social media, each annotated with a human-written implicit narrative explanation. Unlike conventional video understanding tasks focused on recognition or description, we present a benchmark that targets contextual multimodal reasoning. We evaluate current video-language models from two perspectives: explanation, where models must explain the pragmatic comprehension of a video in natural language; and representation, where we adapt reasoning-as-retrieval to test whether model representations align videos with their corresponding implicit narratives in both video-to-text and text-to-video retrieval. Our benchmark provides a diagnostic setting for measuring the gap between multimodal perception and pragmatic comprehension, asking whether current models can move beyond describing what is shown to inferring what is meant.

cs.CL↗
arXiv2026

Ultra-low-loss Ta2O5 waveguide for integrated nonlinear photonics

Integrated nonlinear photonics demands waveguides that simultaneously offer ultra-low-loss, relatively strong nonlinearity, and tight mode confinement. Tantalum pentoxide (Ta2O5) is attractive for this purpose because it combines a broad transparency window, relatively strong third-order nonlinearity, and favorable thick-film growth characteristics. In tightly confined Ta2O5 waveguides, however, enhanced interaction between the optical mode and etched sidewalls makes scattering loss an important limitation for achieving high Q anomalous-dispersion resonators. Here, we mitigate this confinement-scattering trade-off in pure Ta2O5 waveguides using a fabrication flow based on the Damascene concept. The resulting microring resonators (MRRs) with a cross-section of 1.6 μm x 0.85 μm exhibit intrinsic quality factors exceeding 4.0 x 106, corresponding to propagation losses below 10 dB/m, while maintaining anomalous dispersion suitable for bright dissipative Kerr soliton operation. Using these resonators, we generate a coherent bright single-soliton microcomb with a repetition rate of 108.32 GHz. Under optimized dual-pump operating conditions, the platform further exhibits an experimentally accessible single-soliton window spanning 7.5 GHz, indicating a large detuning tolerance for soliton operation; this broad operating window is consistent with the relatively weak thermo-optic response of Ta2O5. Thermal-bistability analysis further indicates that the residual scattering loss is comparable to the material-absorption loss, rather than dominating the loss budget. These results demonstrate that low optical loss, tight mode confinement, and dispersion engineering suitable for soliton formation can be jointly achieved in a pure Ta2O5 platform, providing an ultra-low-loss platform foundation for integrated nonlinear photonic devices and more complex chip-scale nonlinear photonic systems.

physics.optics↗
arXiv2026

Dual-Stream Memory for Vision-Language-Action Policies

Memory-dependent manipulation requires a policy to track execution state and recover visual evidence that is no longer observable. We propose dual-stream memory for vision-language-action policies, assigning these requirements to complementary representations. A task-agnostic Gated DeltaNet compresses history into one temporal token, supervised by progress prediction and action-conditioned feature reconstruction. A task-conditioned retriever retains frames above a fixed relevance threshold; linear multi-scale fusion encodes each frame into sixteen spatial tokens. We then freeze the temporal compressor and jointly optimize retrieval and policy learning. Across sixteen RoboMME tasks, the complete model achieves 49.22% mean success versus 44.51% for the published FrameSamp+Modul reference, improving fourteen tasks and all four suite means. On four real-robot tasks, mean success reaches 62.5% versus 37.5%, with the largest gains on repeated actions requiring execution-stage discrimination. These results support combining execution context with selective visual access, while failures under cup swaps expose persistent identity tracking as an unresolved requirement.

cs.RO↗
arXiv2026

DORS: Dynamic Attention Routing for Diffusion-based Object Removal in Dense Scenes

Object removal aims to eliminate target objects specified by a mask while preserving visual consistency with the surrounding regions. Existing methods typically rely on contextual information from surrounding regions. However, in dense scenes where the surrounding regions contain instances visually similar to the removal target, such reliance often leads to semantic interference, resulting in incomplete removal. This problem arises from erroneous information propagation in the attention space, where masked queries tend to align with such instances due to global similarity matching in self-attention. To address this challenge, we propose a Diffusion-based Object Removal framework for dense Scenes, dubbed DORS, built upon a Dynamic Attention Routing mechanism comprising two complementary components: Instance-Filtered Attention (IFA), which suppresses misleading semantic information from similar instances through dynamically constructed mask-guided attention constraints, and Context-Guided Routing (CGR), which dynamically routes complementary scene information to maintain visual consistency. We further introduce DOR-Bench, a benchmark tailored for object removal in dense scenes. Extensive experiments demonstrate that DORS outperforms state-of-the-art methods, particularly in reducing incomplete removal and duplicate artifacts. The code will be available at https://github.com/httang1224/DORS.

cs.CV↗
arXiv2026

Robust quantum key distribution with imperfections in state preparation and characterization

Quantum key distribution (QKD) offers an approach for secure communication over quantum channels, based on the principles of quantum mechanics. However, the inevitable imperfections in QKD systems can introduce critical vulnerabilities, posing significant challenges to the deployment of QKD in terms of both practical security and performance. In this work, we present a comprehensive security analysis of QKD in the practical scenarios involving imperfections in modulation spaces. Specifically, our analysis focuses on the imperfections related to encoding, intensity modulation, and their characterization. We provide a unified framework based on the operator dominance method to overcome these issues and achieve a tight estimation of the secret key rate (SKR). The simulation results show that the SKR in practical scenarios, with parameters characterized from commercial QKD systems, closely matches the SKR under ideal conditions. In the finite case, the maximum transmission distance can reach up to 250 km with total pulse number $N=10^{12}$ and the SKR closely approaches the asymptotic limit up to 250 km with $N=10^{14}$. This validates the robustness of our method against imperfections in state preparation and characterization. Furthermore, the feasibility and robustness of method in measurement-device-independent (MDI) QKD have also been demonstrated. Our analysis not only promotes advancements in enhancing practical security and performance but also highlights the feasibility of deploying QKD for actual applications.

quant-ph↗
arXiv2026

Experimental Measurement-Device-Independent Quantum Key distribution with Flawed State-Preparation over 300 km

Quantum key distribution (QKD) promises theoretically secure communication. However, it encounters challenges in implementation security and performance due to inevitable device imperfections. Since the proposal of measurement-device-independent (MDI) QKD, the critical step toward practical security is to secure QKD with imperfect sources. The source imperfections manifest as state-preparation uncertainty (SPU) in various aspects, e.g., encoding uncertainty, intensity fluctuation, and imperfect vacuum states. Here, we perform an MDI-QKD experiment and achieve both high practical security and superior performance. We address the general form of SPU and guarantee a tight estimation of the secret key rate based on the operator dominance method. We achieve secure key distribution over 303.37 km, which not only represents the farthest distance in experiments involving SPU but also considers the most SPU scenarios. Our experimental results represent a significant step toward promoting practical and secure quantum communication.

quant-ph↗
arXiv2026

Masked Feature Encoding for Large-Scale Whole Slide Image Representation

Whole slide image (WSI) analysis in computational pathology follows a multiple instance learning (MIL) pipeline where patch embeddings are extracted independently and aggregated for slide-level prediction, but within-slide variance from staining, scanner, and local texture can overwhelm the discriminative signal. We propose Masked Feature Encoding for Multiple Instance Learning (MFE-MIL), a feature-space masking framework that trains a lightweight MLP adapter jointly with a window-based masked reconstruction branch and a MIL classification head. The two objectives are complementary. Classification guides the adapter to suppress within-slide patch variance, while window-based masked reconstruction provides an auxiliary regularizer for the adapted features without using patch coordinates, coordinate graphs, or segmentation preprocessing. The raster patch-extraction order is used only as a weak implicit prior. At inference, the decoder is removed, leaving only the adapter and MIL head. Across CAMELYON16/17, PANDA, and TCGA-BRCA with four diverse encoders, MFE-MIL improves ACC/F1 for nearly all tested aggregator-encoder settings and AUC in most, outperforms coordinate-based spatial methods (CAMIL), and achieves higher AUC than 2DMamba on three of four datasets (UNI). On five TCGA survival cohorts it improves the average concordance index for every aggregator tested, its most consistent gain. Code is available at https://github.com/AtlasAnalyticsLab/MFE-MIL.

cs.CV↗
arXiv2026

Visualizing Three-Dimensional Quantum Fluctuations of Ground-State C$_3$H$_4$ via Deep Sub-Angstrom Coulomb Explosion Imaging

Coulomb explosion imaging (CEI) is a powerful tool for determining static geometries of isolated gas-phase molecules. In this work, taking allene and propyne as prototype, we demonstrate sub-angstrom resolution achieved with highly charged ion-induced CEI. Using 41.9 keV/u Xe$^{20+}$ beam, minute variations of CC single, double and triple bond lengths are clearly visualized, indicating that an ultrahigh resolution of 0.1 $\mathring{\mathrm{A}}$ is achieved. This unprecedented resolution not only makes it possible to distinguish CC bonds with different bond orders, but also lays a solid foundation for direct observation of quantum fluctuations inherent to 3D structure of a ground-state molecule. Notably, the coupled collective motion of the two CH$_2$ moieties spaced by a central carbon within allene is unambiguously visualized in our measurements.

physics.chem-ph↗
arXiv2026

Disentangling Dual Image References in Frequency Aware Diffusion Models for Personalized Generation

Personalized image generation aims to synthesize text-driven images conditioned on reference images, while mainly casting the generation as image customization for foreground and style transfer for background. Previous arts of diffusion models suffers from the text misalignment with background for image customization and foreground for style transfer during the denoising process. Such facts, as we observed, rooted from the entanglement among hybrid frequency bands during the denoising process. To address such salient limitation, in this paper, we study personalized generation based on dual references - customization and color and style reference - and propose a paradigm to disentangle these Dual image references within Frequency-aware Diffusion Models, dubbed Dual-FDM, to simultaneously tackle two crucial personalized image generation tasks: customization style transfer and color style transfer, by disentangling different frequency bands via mask strategy within frequency domain. For customization style transfer, we replace the mid-frequency band of the background in the style reference with that from the foreground of the customized reference. For color style transfer, we substitute the low-frequency band of the background in the style reference with that from both the foreground and background of the color reference. Both the substituted frequency bands are used as the key and value to reconstruct the query foreground and background of the denoised personalized image.Extensive experiments validate the superiority of Dual-FDM over the state-of-the-art diffusion models for personalized image generation. Our code can be accessed from https://github.com/htyjers/Dual-FDM.

cs.CV↗
arXiv2026

Systemization of Knowledge (SoK): Human-Centered AI Safety for Youth

While HCI increasingly examines AI-safety for youth, the literature lacks a comprehensive view of what risks have been identified, how they are addressed, and whether proposed protections work in-practice. We systematically reviewed 100 empirical HCI studies involving children and youth interacting with or exposed to AI across schools, homes, care settings, and public services. Using the YAIR taxonomy for risks and the MIT Mitigation Taxonomy for countermeasures, we map which risks have been identified, whether each risk is addressed by countermeasure(s), and whether each countermeasure for that risk is implemented and even evaluated. The risk-countermeasure mapping shows that most risks are matched only with proposed/ideated countermeasures; few countermeasures have been implemented, and fewer still evaluated; and existing evaluations often measure technical performance rather than protection from harm. We identify where coverage is absent, where safeguards remain untested, and propose concrete directions for HCI research to strengthen youth AI-safety.

cs.HC↗
arXiv2026

JudgeMoE: Distributional Aggregation for LLM-as-a-Judge

When an LLM judge scores an output, its score distribution retains uncertainty and disagreement information that is lost after scalar compression. We introduce JudgeMoE, a lightweight aggregator that assigns example-specific weights to cached judge score distributions and fuses them before computing a final score. A protocol study shows that score-range choice is unstable across judge--dataset settings and that soft scoring usually outperforms hard decoding. On the original 10-cell benchmark, JudgeMoE improves mean Spearman over uniform log pooling by $+0.079$. Applying the same configuration to six additional cells yields a $+0.0393$ mean gain over the strongest local single judge across 16 cells, with positive differences in 12/16 cells and a one-sided Wilcoxon signed-rank $p=0.0091$. Validation-based analyses further show that the preferred aggregation method depends on the task and judge pool.

cs.CL↗
arXiv2026

Towards Trustworthy Physical Intelligence: From Theory to Practice Across Life Cycle

Physical intelligence refers to intelligence systems that understand, reason about, and act in accordance with the physical world and its underlying laws, dynamics, and constraints. Unlike conventional AI systems, physical intelligence interacts continuously with uncertain physical environments, and its actions produce consequences that are physically irreversible. As existing trustworthy AI frameworks have been developed primarily for digital AI systems, they do not fully capture the distinctive challenges of Physical Intelligence, such as physical safety, cyber-physical security, and physical manufacturing process. To address this gap, we present a survey of trustworthy physical intelligence principles. First, we characterize the core capabilities and challenges of physical intelligence. Second, we examine the role of physics in AI. Third, we trace the end-to-end physical intelligence life cycle across five core stages and introduce Trustworthy Physical Intelligence Operationalization (T-PAIO). Fourth, we develop the Trustworthy Physical Intelligence (T-PAI) framework, a theoretical framework that organizes key trustworthiness principles and provides a foundation for governing trustworthy physical intelligence systems.

cs.AI↗
arXiv2026

World Embedding Benchmark

Physical fidelity has received increasing attention in world models and video generation, yet how video representations encode physical information remains less understood. We introduce the World Embedding Benchmark, comprising 8,000 controlled simulation cases from 80 families spanning fluid mechanics, solid mechanics, dynamics, and optics & electromagnetism. Each case pairs a rendered video with simulation-derived physical annotations, supporting three complementary tasks: text-video retrieval, physical-property regression, and multiple-choice video-description pair classification. We use these tasks to distinguish cross-modal physical alignment from the recoverability of quantitative physical information. Evaluated pre-trained omnimodal embedding models show weak retrieval and near-chance within-family pair classification, while lightweight probes recover useful physical information from frozen video embeddings. Continual contrastive training with physics-specific video-text pairs improves retrieval and pair classification but degrades physical-property regression, revealing a trade-off between alignment and quantitative information recoverability. Finally, we use the embeddings to retrieve reference videos for retrieval-augmented generation with MiniMax-H3. Retrieved references improve the physical fidelity of generated videos, with stronger retrieval models yielding larger gains in our experiments. Together, these findings highlight the need to evaluate physical alignment and property recoverability jointly, and demonstrate the utility of physical representations for improving video generation.

cs.CV↗
arXiv2026

Approval Laundering: Systematizing Approval--Execution Binding Failures in AI Coding-Agent Harnesses

Modern AI coding-agent harnesses (Claude Code, Codex CLI, Cursor) rest their security boundary on a largely unexamined assumption: that the action A a human approves is the same action A' the harness executes, where A is fixed by a stated policy for what a scope grant or session-scoped approval authorizes. We show this assumption fails systematically and reproducibly. We introduce Approval Laundering, a taxonomy of six failure modes by which a harness's enforcement mechanism silently substitutes A' for A after approval: Scope, Argument, Temporal, Tool, Delegation, and Semantic laundering. Unlike prior work that evaluates risk classifiers against static corpora or infers implicit authorization boundaries, we study credential-binding integrity: given an already-approved action, does the harness dispatch exactly that action? Instrumenting Claude Code's pre-execution mediation point (PreToolUse), we conduct a controlled, headless, repeated-measures study of all six classes (N=19-20 runs each), reporting a Bound-Gap Rate (BGR) with Wilson confidence intervals and inter-rater agreement (kappa=1.0). We prototype Approval Token, a keyed capability Hk(principal, agent_id, session_id, tool, arguments, scope, expiry) issued by a mediator that never returns the key to the agent, evaluated via paired before/after replay of 118 runs (McNemar's exact test). The token fully eliminates Delegation laundering and, for our seeded session-identity-mismatch construction, Temporal laundering (p<10^-5), but by design leaves Scope laundering unaffected and shows no significant reduction in Argument laundering (p=1): an honest negative result, since these two classes leave every recorded dispatch field unchanged, diverging one process level below what a field-only verifier can observe. We discuss implications for defenses that bind only at the tool-invocation boundary.

cs.CR↗
arXiv2026

Bit-Accurate Modeling of GPU Matrix Multiply-Accumulate Units: Demystifying Numerical Discrepancy and Accuracy

Modern AI accelerators rely on matrix multiply-accumulate units (MMAUs), such as NVIDIA Tensor Cores and AMD Matrix Cores, to accelerate deep neural network workloads. MMAUs expose only instruction-level or API-level interfaces of matrix multiply-accumulate (MMA) operations, while leaving internal floating-point arithmetic behavior undocumented. Consequently, MMAUs across vendors and architectural generations often produce numerical discrepancies for identical inputs, and sometimes exhibit reduced numerical accuracy that can cause training instability. Diagnosing and understanding the root causes of these effects is challenging without white-box models of their arithmetic behavior. This paper proposes closed-loop feature probing (CLFP), a generic and systematic framework for constructing bit-accurate arithmetic behavior models of MMA operations. Based on this framework, we analyze all MMA instructions on ten GPU architectures spanning NVIDIA Volta through RTX Blackwell and AMD CDNA1 through CDNA3, and derive the first bit-accurate arithmetic models for these MMAUs. Our models explain previously observed cross-platform numerical discrepancies and accuracy issues, enable white-box numerical error analysis, reveal four types of precision bottlenecks and one type of numerical asymmetry, and inform software workarounds as well as design suggestions for future MMAUs. This work is open-source at https://github.com/microsoft/MMA-Sim

cs.AR↗
arXiv2026

One Rollout Is All You Get: Fully Test-Time Adaptation for GUI Agents

GUI agents are deployed with frozen weights and discard everything they experience on the job. Existing ways to update an agent's weights assume something deployment withholds: ground truth, rollouts beyond the single attempt (retries, samples, practice runs), or a learning phase other than deployment. Because GUI actions can be irreversible, a deployed agent gets one attempt per task occurrence, in arrival order, and every attempt counts. No ground truth is available at any point. We define fully test-time adaptation for GUI agents by these constraints and pair it with a minimal weight-space method, SOLO. Auxiliary models read each episode: a judge selects the episodes it deems successful, and a proposer-verifier pair relabels a failed episode's prefix with the subtask that prefix completed. Admitted episodes enter a short sliding window, and each admission updates a small adapter by top-K self-distillation on the agent's own predictions, provided the window holds a judged success. On recurring task streams built from WebArena, VisualWebArena and MobileWorld, SOLO improves on the frozen agent with both UI-TARS-7B and Qwen3-VL-8B, by three to six points of success rate, and exceeds two in-setting memory methods on the web streams.

cs.LG↗
arXiv2026

Remember by Asking: Retrieval-Induced Memory Evolution for LLM Agents

Long-term memory is essential for language agents to maintain coherent and effective behavior over extended, multi-session interactions. Existing memory systems mainly use retrieval at read time, while write-time memory formation still relies on direct extraction or compression. However, when future information needs are unknown, compressing an entire interaction in one pass can overlook locally important details that may matter later. To this end, we introduce RIME, a retrieval-induced memory framework that shifts memory construction from monolithic compression toward evidence-centered integration. RIME uses generic self-questions to retrieve focused dialogue evidence and grounds memory formation in both the retrieved evidence and relevant historical memories, which are jointly reconciled into an evolving memory bank with temporal and provenance information. At inference time, compressed memory serves as the primary rather than the sole source of evidence: when it cannot support an answer, RIME retrieves relevant source dialogue together with its local context to recover information omitted during memory formation, without resorting to full-history processing. Extensive experiments on LoCoMo with Qwen3-235B-A22B and GPT-5.6 Sol show that RIME consistently achieves the best performance across all three quality metrics among the compared methods, while requiring substantially fewer query-time LLM tokens.

cs.CL↗