Searcharxiv⌕ Search

arXiv subjects

Bowen Wang

Publications and source records attributed to Bowen Wang.

At least 19 recordsLinked to original sources

Learning to Edit IFC Building Models: Verifiable Checkers as the Training Signal for Dependency-Aware Editing Agents

Editing a building information model in the Industry Foundation Classes (IFC) schema requires propagating changes to dependent elements while preserving unrelated entities. Existing language-model agents for IFC editing use prompted models without learning from edit outcomes. Target-file comparisons do not specify which alternative edits are allowed or which unrelated entities must remain unchanged. We present IfcGym, a CodeAct environment in which a language model edits IFC through executable code and four checkers verify validity relative to the input model, intent, dependency and preservation against a task specification. Their conjunction is strict success. IfcEdit distils verified teacher episodes into a Qwen3.5-9B student through supervised fine-tuning (SFT), then applies Group Relative Policy Optimization (GRPO) with strict success as the reward. On IfcEditBench, 1,990 editing tasks on 13 held-out real projects, the final student reaches 86.78% strict success, from 79.15% after SFT. The paired confidence interval for the difference from its teacher includes zero; the same recipe on a 27B model reaches 91.76% and exceeds the teacher. On the public benchmarks BIM-Edit and BIBIMBAP the final student scores 55.19 and 54.50. On IfcEditBench, the students outperform all prompted models except the teacher under matched tool access; both evaluated frameworks also score lower. Rewarding all four checkers yields 4.83 points more strict success than rewarding intent and validity alone.

cs.AI↗

Capacity-Constrained Delayed Feedback for Online Conformal Risk Control

Online conformal risk control (OCRC) enables reliable uncertainty quantification in machine learning by converting model outputs into prediction sets with long-term risk guarantees. These guarantees, however, hinge on the assumption that the loss of each round is revealed immediately and losslessly. In practice, feedback may be delayed, and limited resources may prevent tracking all pending predictions, causing some feedback to be lost. To address this challenge, we develop capacity-aware OCRC (CA-OCRC), a framework that jointly determines how to allocate limited tracking feedback capacity and how to construct prediction sets. CA-OCRC is based on a randomized tracking policy that samples a proxy delay to decide whether, and for how long, feedback for the current round is tracked. CA-OCRC updates the prediction sets via inverse probability weighting, compensating for the possibility of feedback lost due to delays and limited tracking capacity. We analyze risk bounds for both arbitrary and stochastic delays, revealing a trade-off between receiving more feedback and avoiding capacity saturation. Building on these bounds, we optimize the proxy delay distribution, showing scaling laws for the tracking capacity that ensure the same long-term risk as OCRC with unlimited feedback. Experiments on medical imaging, natural language processing, computer vision, and wireless communications show the effectiveness of CA-OCRC as compared to existing OCRC mechanisms.

cs.LG↗

Drop the Details, Get the Gist: Dynamic Conceptual Representation for Video Recognition

Concept Bottleneck Models (CBMs), which route predictions through a layer of human-interpretable concepts, have substantially improved the interpretability of visual classification by exposing an explicit, language-based rationale for each prediction. Most existing work focuses on static images, and temporal dynamics in videos have been underexplored in conceptual representation. We propose a framework that extends concept-based image recognition to procedural video understanding. It represents a multi-step video as a temporal sequence of segment-level concept score vectors and feeds this sequence into a Transformer encoder to predict the class label. We also explore a way to estimate the contribution of each concept for further interpretability. We evaluated three concept sets of different granularities on cooking and household-activity videos and found that coarser-grained concepts are sufficient for classifying these videos, which implies that fine-grained textual descriptions may be unnecessary.

cs.CV↗

KnowMT-Bench: Benchmarking Knowledge-Intensive Long-Form Question Answering in Multi-Turn Dialogues

Multi-Turn Long-Form Question Answering (MT-LFQA) is a key application paradigm of Large Language Models (LLMs) in knowledge-intensive domains. However, existing benchmarks are limited to single-turn dialogue, while multi-turn dialogue benchmarks typically assess other orthogonal capabilities rather than knowledge-intensive factuality. To bridge this critical gap, we introduce \textbf{KnowMT-Bench}, a new benchmark designed to systematically evaluate MT-LFQA for LLMs across knowledge-intensive fields, including medicine, finance, and law. To faithfully assess the model's real-world performance, KnowMT-Bench employs a dynamic evaluation setting where models generate their own multi-turn dialogue histories given logically progressive question sequences. Our experiments on a diverse suite of LLMs show a clear degradation in both factual capability and information delivery efficiency in multi-turn dialogue. We further probe the underlying causes and investigate mitigation strategies spanning domain-specific finetuning, active context denoising, and RAG. The results show domain-dependent gains from finetuning, while active context denoising and RAG more consistently mitigate these issues, with RAG notably reversing this performance degradation. These findings underscore the importance of our benchmark for evaluating LLMs' conversational factual capabilities in real-world applications. Code and data are publicly available.\footnote{\href{https://github.com/Sven-LI-sankyuu/KnowMT-Bench}{\texttt{https://github.com/Sven-LI-sankyuu/KnowMT-Bench}}}

cs.CL↗

Leveraging LLM-Generated Explanations for Detecting Emotionally Rewritten Fake News

The spread of fake news may cause severe social consequences. Existing fake news detection methods mainly focus on stylistic variations or incorporate external information such as explanations. However, news articles are often rewritten under different emotional backgrounds while preserving their underlying factual claims, which may affect the robustness of detection models. In this work, we investigate fake news detec- tion under fact-preserving emotional variations. To study this problem, we construct emotion-rewritten test sets and generate explanations from the original news articles as stable background knowledge. We then propose a Gated Cross Attention (GCA) framework that adaptively integrates emotionally rewritten news with the corresponding explanations, enabling the model to focus on informative explanation content while reducing potential mismatches caused by emotional reframing. Experiments on PolitiFact, GossipCop, and LUN demonstrate that the proposed method achieves notable improvements under multiple emotional conditions on PolitiFact and LUN, while maintaining competitive performance on GossipCop. We further analyze the effects of explanation guidance and gating mechanisms under different emotional conditions. Our code and data are available at: https://github.com/Flulike/fakenews gca .

cs.CL↗

SURE-EVAL: A Systematic and Unified Agentic Framework for Reproducible Evaluation

Audio and speech models are released rapidly, but reported scores often conflate model capability with deployment and evaluation choices. The same checkpoint can produce different predictions under different runtimes, hardware, decoding settings, or fallback policies. Even fixed predictions can receive different scores under different normalization and metric implementations. Existing speech benchmarks standardize selected datasets or scoring procedures, but rarely connect heterogeneous model onboarding, controlled inference, and versioned scoring in one executable workflow. We introduce SURE-EVAL, a Systematic and Unified Agentic framework for Reproducible Evaluation of audio and speech systems. A Tool Agent Workflow converts model releases into isolated, verified callable tools. A Main Agent Workflow commits task-specific inference and scoring protocols, then delegates all score-bearing operations to versioned deterministic programs. Each result retains its runtime, protocol, pipeline nodes, predictions, and audit artifacts. Across 18 public releases covering automatic speech recognition, text-to-speech, voice conversion, speaker diarization, speaker-attributed recognition, and multi-task audio understanding, a Codex-only baseline completes 12 models in one shot, while the same agent with the SURE-EVAL Tool Agent Workflow completes all 18. We also conduct unified evaluations over seven ASR test conditions and two TTS subsets. A protocol analysis of three TTS systems finds absolute differences of 0.02-0.52 points between paper-reported and unified results, with the direction varying by model and language. These results show that reproducible evaluation requires controlling both model execution and output scoring.

eess.AS↗

Experimental certification of multipartite Bell correlations using only few-body symmetric correlations

Certifying Bell correlations in increasingly larger multipartite quantum systems is challenging because conventional Bell inequalities rely on many-body correlations, which become experimentally demanding to estimate as system size grows. Here we demonstrate that multipartite Bell correlations can be certified using few-body permutation-invariant (PI) correlations with a fixed maximal correlation order. We experimentally test few-body PI Bell correlation witnesses tailored for six- and eight-photon postselected Dicke states. The addition of a few four-body correlators is found to significantly enhance the noise tolerance compared with previously studied two-body witnesses: for instance, for six qubits, the selected witness tolerates a white-noise fraction of $p_c\approx0.301$, compared with $p_c\approx0.048$ for the best two-body PI inequality with the same two settings per party. For six and eight photons respectively, we observe violations of the witness bounds by $32$ and $8.9$ standard deviations, arising from postselected polarisation states with fidelities of $0.933(4)$ and $0.916(11)$ to the corresponding half-excitation Dicke states. The witnesses for the six- and eight-photon cases use the same symmetric two- and four-body correlators, just with different coefficients. These results demonstrate an experimentally accessible approach for certifying multipartite Bell correlations in larger systems without requiring full-body correlation measurements.

quant-ph↗

Loss-Resilient Quantum Networking with Dicke Entanglement

Multipartite entanglement can endow a quantum network with capabilities unavailable classically, yet their evolution under progressively reduced physical and informational access remains largely unexplored. Here we realize the first eight-photon half-excitation Dicke state, with a fidelity of 0.922(11), and use it to implement two distinct network tasks: open-destination teleportation and quantum telecloning. An unknown state can be teleported from one agent to any of seven remote agents with fidelities of 0.926(7)-0.941(6), while seven approximate clones are distributed with a fidelity of 0.713(3), close to the optimal value of 5/7. The protocol remains above the classical limit under imperfect photon detection, transmission loss and missing information from participating agents, and outperforms the corresponding GHZ benchmarks. As fewer agents contribute, the teleportation fidelity falls through a stepwise hierarchy whose two extremes recover open-destination teleportation and optimal telecloning, connecting both tasks within a single operational framework. These results establish Dicke entanglement as a promising architecture for loss-resilient quantum networks, opening a route towards quantum technologies that retain useful functionality under realistic operating conditions.

quant-ph↗

ANCHOR: An External LLM-Driven Supervisory Module Facilitating Healthy Evolution in Self-Evolving Systems

Self-evolving agents improve through continual self-play and self-generated learning signals, but their internally generated tasks and verifier signals provide limited coverage of phase-level errors, allowing capability degradation and safety drift to accumulate. We introduce ANCHOR, an LLM-based supervisory framework that delivers evaluative feedback at multiple phases of self-evolution and aggregates reviewed signals into context for subsequent steps. We retrofit two representative open-source self-evolving agent frameworks with ANCHOR, and evaluate them across coding, mathematical reasoning, and safety. Our results show that ANCHOR substantially improves safety performance while maintaining stable performance on the core capabilities of the underlying self-evolving agents. Further analyses provide practical insights for future research, showing that execution-result-based supervision is particularly effective and that increasing supervision frequency yields diminishing returns. Together, these results support external LLM-based supervision as a practical approach to developing safer, more stable, and controllable self-evolving agent systems.

cs.AI↗

BRF-GS: Hyperspectral Bidirectional Reflectance Factor Modeling and Image Generation Based on 3D Gaussian Splatting

The bidirectional reflectance factor (BRF) characterizes the directional radiative properties of terrestrial surfaces. However, existing three-dimensional (3D) radiative transfer models require complex scene construction and computationally intensive radiative transfer solvers, limiting efficient generation of multi-angle hyperspectral reflectance imagery. 3D Gaussian Splatting (3DGS) offers an efficient framework for neural scene representation and novel view synthesis, but its low-order spherical harmonics representation is insufficient for complex directional reflectance, while the high dimensionality and inter-band quality differences of hyperspectral data introduce additional challenges. To address these challenges, we propose BRF-GS, a 3DGS-based framework for BRF modeling and hyperspectral reflectance image generation. BRF-GS introduces a hybrid BRDF-driven kernel to represent complex directional reflectance, selects geometry-reliable spectral bands for robust 3D scene initialization, and adopts a two-stage training strategy that decouples geometry optimization from spectral modeling. We further construct the AIR-BRF dataset, a multi-angle hyperspectral directional reflectance dataset comprising three scenes with diverse natural and artificial targets. Experiments demonstrate that BRF-GS achieves superior spatial and spectral fidelity and accurately reproduces characteristic view-dependent BRF responses. The proposed framework provides an efficient data-driven approach for BRF modeling and multi-angle hyperspectral reflectance image generation in remote sensing scenes.

cs.CV↗

The Blending Ratio Is Not Where the Performance Is: Diagnosing Prototype Blending for Few-Shot Adaptation of Vision-Language Models

Many few-shot adaptation methods for vision-language models classify with a convex combination of the zero-shot text prototype and the mean of the K labelled image features, with a single blending ratio routinely tuned on held-out labels, often on the test set itself. We ask what the family's own bias-variance justification invites: what is the right ratio, can it be estimated without validation data, and is finding it where the performance is? First, the ratio minimising prototype mean-squared error has a closed form whose support-set plug-in is exactly a positive-part James-Stein coefficient shrinking towards the text prototype. Across 4,800 cells (ten datasets, five backbones including SigLIP, five shot counts, five seeds, four prompt tiers) this theoretically optimal ratio is a reliable estimate of the wrong quantity: on the 950 primary-tier cells where it is defined it trails a test-set-oracle ratio by 8.5 points. It saturates near 1, discarding the text prior for a nearest-class-mean classifier, because 78% of the text-image prototype distance it treats as bias is a class-independent offset that the arg max largely cancels. We prove the mechanism and bound its share of the damage at 26% by a counterfactual. Second, leave-one-out on the support set alone sets a ratio landing within 0.9 points of the oracle blend, so it is estimable without validation data. Third, validation-free linear probes beat even the oracle-tuned blend: CLAP by +1.9 points and LP++ by +1.5 on average, and at K >= 4 all four validation-free baselines sit above the oracle, the linear probes by margins excluding zero. These results locate the ceiling in the model class, not the hyperparameter: the ratio can be set near-optimally for free, and it is still not where the performance is. Code, cached features, per-cell records: https://huggingface.co/datasets/Liangzhi-Li/clipbench-blending

cs.CV↗

Investigation of Air Fluidization during Intruder Penetration in Sand

Self-burrowing robots navigating through granular media benefit from airflow-assisted burrowing, which reduces penetration resistance. However, the mechanisms underlying airflow-granular interactions remain poorly understood. To address this knowledge gap, we employ a coupled computational fluid dynamics and discrete element method (CFD-DEM) approach, supplemented by experimental cone penetration tests (CPT) under varying airflow conditions, to investigate the effects of aeration on penetration resistance. Experimental results reveal a nonlinear relationship between penetration resistance reduction and depth, wherein resistance approaches near-zero values up to a critical depth, beyond which the effectiveness of fluidization diminishes. Simulations demonstrate that higher airflow rates enhance the mobilization of overlying grains, increasing the critical depth. A detailed meso- and micro-scale analysis of particle motion, contact forces, and fluid pressure fields reveals four distinct penetration stages: particle ejection and channel formation, channel sealing, channel refill, and final compaction. These findings contribute to a deeper understanding of granular aeration mechanisms and their implications for geotechnical engineering, excavation technologies, and the development of self-burrowing robotic systems.

cond-mat.soft↗

Qwen-CUA: Native Computer Use for (almost) Everything

Native computer use offers a general interface for agents to operate almost any software available to people, but requires long-horizon state tracking, large-scale interactive experience, and learning from sparse yet verifiable outcomes. We introduce Qwen-CUA, a native computer-use agent with a 397B-A17B Qwen mixture-of-experts backbone. It observes only screenshots and acts through keyboard and mouse events, without DOM trees, accessibility metadata, or task-specific APIs. Its scaffold maintains up to 20 active screenshots and folds older visual history in fixed-size blocks to retain recent evidence while preserving reusable prompt prefixes. For training, we build a cloud rollout fleet with access to nearly 100,000 vCPUs and tens of thousands of concurrent environments, construct approximately 40,000 verifiable tasks, and collect personalized long-horizon workflows across everyday and professional software. We optimize complete trajectories with verifiable rewards and trajectory slicing, while iterative training runs refresh supervised data and recalibrate reinforcement-learning tasks. Across eight benchmarks, Qwen-CUA outperforms Qwen3.7 and remains competitive with leading proprietary systems, reaching 86.2 on OSWorld-Verified and 18.5/48.4 binary/partial completion on OSWorld 2.0. Scaling the same recipe to a model with over one trillion parameters yields Qwen-CUA-Max, improving these scores to 87.6 and 21.2/53.3. Qwen-CUA also reduces RedTeamCUA attack success from 36.6 to 16.4 relative to Qwen3.7. Efficiency analyses, a browser deployment, and Bash-augmented experiments further characterize practical behavior. These results establish native computer use as a broadly capable agent foundation and highlight scalable verifiable interaction and hybrid tool use as key directions.

cs.LG↗

DS@GT ARC at ImageCLEFmedical 2026: Architectural Diversity for Concept Detection and Foundation-Model Scaling for Caption Prediction in Medical Image Analysis

We describe the DS@GT submissions to the ImageCLEFmedical Caption 2026 challenge, which continues a long-running benchmark on the ROCOv2 dataset with two tracks: Concept Detection (Task 1), assigning UMLS Concept Unique Identifiers (CUIs) to radiology images, and Caption Prediction (Task 2), generating natural-language captions. For Task 1, our primary submission was a three-way late-fusion ensemble of ConvNeXt-V2, BiomedCLIP ViT-B/16, and DenseNet-169 with a regularized ''Honest Threshold Tuning'' procedure designed to avoid validation overfitting on rare concepts; this submission ranked first on the official submission with a primary $F_1$ of $0.5790$ and a secondary $F_1$ of $0.9657$. In parallel, we submitted a training-free KNN retrieval pipeline over frozen BiomedCLIP embeddings, which reached a primary $F_1$ of $0.5780$ and a secondary $F_1$ of $0.9599$-essentially matching the fine-tuned ensemble on the primary track at a fraction of the cost. For Task 2, our submissions included a fine-tuned Gemma-3 27B model (overall $0.3571$, ranking third in the official submission), a fully fine-tuned BLIP pipeline with custom Vizwins merging ($0.3564$), and a zero-shot MedGemma-4B run with a PubMed-style prompt ($0.3186$), spanning a wide range of model scales and training costs. Code: https://github.com/dsgt-arc/imageclef-caption-2026.

cs.CV↗

Design in Tiles: Automating GEMM Deployment on Tile-Based Many-PE Accelerators

Tile-based many-Processing Element (PE) accelerators can achieve competitive performance on General Matrix Multiplication (GEMM), but they are extremely hard to program, as their optimal software mapping is deeply coupled with hardware design which is unwieldy to manual deployment. We propose "Design in Tiles (DiT)", an automated framework connecting a deployment toolchain with a configurable executable model for these accelerators. For evaluation, we apply our framework to GEMM targeting a large acceleration configuration (e.g., 32x32 tiles, 1979 TFLOPS@FP8, 4 TB/s Bandwidth) comparable to an NVIDIA GH200. We achieve higher PE utilization than GH200 with its expert-tuned GEMM libraries, achieving 1.2-2.0x speedup across diverse matrix shapes.

cs.DC↗

At-the-Roofline Sparse Tensor Contractions on Vector Processors for Transformer Inference

Fine-grained weight pruning and activation sparsification have emerged as effective approaches for reducing the compute and memory cost of inference for Transformer models. In the moderate-sparsity regime, Gustavson's dataflow provides a natural execution model for exploiting both activation and weight sparsity on vector processors through metadata-driven indexed accumulation. However, existing RVV architectures lack native support for this pattern, forcing kernels to rely on software index decoding and L1-backed indexed memory operations that keep sparse tensor contractions far below their roofline performance bound. We present Ventaglio, a runtime-configurable sparse execution unit coupled with RVV ISA extensions that drives sparse tensor contractions toward their roofline through indexed gather-accumulate-scatter support. Integrated into an open-source vector processing cluster and implemented in 12nm FinFET, Ventaglio accelerates sparse tensor contraction kernels by $6.9\text{--}7.4\times$ over optimized RVV baselines, with only $3.1\%$ area overhead for a cluster of tightly-L1 coupled vector processing elements. We build a performance-accurate instruction-level model of the Ventaglio extension, calibrate it against RTL implementation, and leverage it for scale-out performance analysis on a large $4\times4$ multi-cluster system. Using a DuoGPT-pruned LLaMA-3-8B model with practical $40\text{--}60\%$ dual sparsity, Ventaglio achieves $2.40\text{--}5.25\times$ and $2.06\text{--}3.16\times$ speedup over dense baselines during prefill and autoregressive decoding, respectively.

cs.AR↗

Sensing Security Oriented OFDM-ISAC Against Multi-Intercept Threats

In recent years, security has emerged as a critical aspect of integrated sensing and communication (ISAC) systems. While significant research has focused on secure communications, particularly in ensuring physical layer security, the issue of sensing security has received comparatively less attention. This paper addresses the sensing security problem in ISAC, particularly under the threat of multi-intercept adversaries. We consider a realistic scenario in which the sensing target is an advanced electronic reconnaissance aircraft capable of employing multiple signal interception techniques, such as power detection (PD) and cyclostationary analysis (CA). To evaluate sensing security under such sophisticated threats, we analyze two critical features of the transmitted signal: (i) power distribution and (ii) cyclic spectrum. Further, we introduce a novel ergodic cyclic spectrum metric which leverages the intrinsic mathematical structure of cyclostationary signals to more comprehensively characterize their behavior. Building on this analysis, we formulate a new ISAC design problem that explicitly considers sensing security, and we develop a low-complexity, efficient optimization approach to solve it. Simulation results demonstrate that the proposed metric is both effective and insightful, and that our ISAC design significantly enhances sensing security performance in the presence of multi-intercept threats.

eess.SP↗

OSWorld 2.0: Benchmarking Computer Use Agents on Long-Horizon Real-World Tasks

Existing computer-use benchmarks fail to capture the realism, complexity, and long-horizon demands of real-world computer use, limiting their ability to reveal the limitations of frontier agents. We introduce OSWorld 2.0, a benchmark of 108 long-horizon computer-use workflows across everyday and professional tasks, designed to capture complex and challenging real-world phenomena. Each task represents a realistic end-to-end workflow that takes human users a median of about 1.6 hours to complete and requires an average of 318 tool calls with Claude Opus 4.7 using maximum thinking, compared with about 30 in OSWorld 1.0. OSWorld 2.0 targets challenge phenomena that are common in real workflows yet underrepresented in prior benchmarks, spanning interaction-design challenges such as streaming interaction and dynamic environments, as well as agent-pattern challenges such as cross-source reasoning, implicit-state inference, and visual-spatial precision. Tasks are grounded in authentic input artifacts and cross-referenced against realistic stateful user profile data, and include separate safety reports auditing safety-sensitive execution. Under our primary binary-completion metric at 500 steps, Claude Opus 4.8 with maximum thinking and batched tool calls scores best but still completes only 20.6% of tasks at a 54.8% partial score; GPT-5.5 is far more token-efficient yet plateaus near 13%. These results show that current agents are still far from professional-level computer use: rather than stumbling on basic GUI control or coding, they lose track of constraints, miss information that arrives mid-task, guess rather than ask the user, and skip verification, struggling most when a task hinges on hidden state they must recover.

cs.AI↗