arXiv subjects

Wenxuan Lu

Publications and source records attributed to Wenxuan Lu.

8 recordsLinked to original sources

arXiv2026

Trans-RAG: Query-Centric Vector Transformation for Secure Cross-Organizational Retrieval

Retrieval Augmented Generation (RAG) systems deployed across organizational boundaries face fundamental tensions between security, accuracy, and efficiency. Current encryption methods expose plaintext during decryption, while federated architectures prevent resource integration and incur substantial overhead. We introduce Trans-RAG, implementing a novel vector space language paradigm where each organization's knowledge exists in a mathematically isolated semantic space. At the core lies vector2Trans, a multi-stage transformation technique that enables queries to dynamically "speak" each organization's vector space "language" through query-centric transformations, eliminating decryption overhead while maintaining native retrieval efficiency. Security evaluations demonstrate near-orthogonal vector spaces with 89.90° angular separation and 99.81% isolation rates. Experiments across 8 retrievers, 3 datasets, and 3 LLMs show minimal accuracy degradation (3.5% decrease in nDCG@10) and significant efficiency improvements over homomorphic encryption.

cs.CR↗

arXiv2026

PRISMA: Reinforcement Learning Guided Two-Stage Policy Optimization in Multi-Agent Architecture for Open-Domain Multi-Hop Question Answering

Answering real-world open-domain multi-hop questions over massive corpora is a critical challenge in Retrieval-Augmented Generation (RAG) systems. Recent research employs reinforcement learning (RL) to end-to-end optimize the retrieval-augmented reasoning process, directly enhancing its capacity to resolve complex queries. However, reliable deployment is hindered by two obstacles. 1) Retrieval Collapse: iterative retrieval over large corpora fails to locate intermediate evidence containing bridge answers without reasoning-guided planning, causing downstream reasoning to collapse. 2) Learning Instability: end-to-end trajectory training suffers from weak credit assignment across reasoning chains and poor error localization across modules, causing overfitting to benchmark-specific heuristics that limit transferability and stability. To address these problems, we propose PRISMA, a decoupled RL-guided framework featuring a Plan-Retrieve-Inspect-Solve-Memoize architecture. PRISMA's strength lies in reasoning-guided collaboration: the Inspector provides reasoning-based feedback to refine the Planner's decomposition and fine-grained retrieval, while enforcing evidence-grounded reasoning in the Solver. We optimize individual agent capabilities via Two-Stage Group Relative Policy Optimization (GRPO). Stage I calibrates the Planner and Solver as specialized experts in planning and reasoning, while Stage II utilizes Observation-Aware Residual Policy Optimization (OARPO) to enhance the Inspector's ability to verify context and trigger targeted recovery. Experiments show that PRISMA achieves state-of-the-art performance on ten benchmarks and can be deployed efficiently in real-world scenarios.

cs.AI↗

arXiv2025

Depth Any Panoramas: A Foundation Model for Panoramic Depth Estimation

In this work, we present a panoramic metric depth foundation model that generalizes across diverse scene distances. We explore a data-in-the-loop paradigm from the view of both data construction and framework design. We collect a large-scale dataset by combining public datasets, high-quality synthetic data from our UE5 simulator and text-to-image models, and real panoramic images from the web. To reduce domain gaps between indoor/outdoor and synthetic/real data, we introduce a three-stage pseudo-label curation pipeline to generate reliable ground truth for unlabeled images. For the model, we adopt DINOv3-Large as the backbone for its strong pre-trained generalization, and introduce a plug-and-play range mask head, sharpness-centric optimization, and geometry-centric optimization to improve robustness to varying distances and enforce geometric consistency across views. Experiments on multiple benchmarks (e.g., Stanford2D3D, Matterport3D, and Deep360) demonstrate strong performance and zero-shot generalization, with particularly robust and stable metric predictions in diverse real-world scenes. The project page can be found at: \href{https://insta360-research-team.github.io/DAP_website/} {https://insta360-research-team.github.io/DAP\_website/}

cs.CV↗

arXiv2025

Cultivating Game Sense for Yourself: Making VLMs Gaming Experts

Developing agents capable of fluid gameplay in first/third-person games without API access remains a critical challenge in Artificial General Intelligence (AGI). Recent efforts leverage Vision Language Models (VLMs) as direct controllers, frequently pausing the game to analyze screens and plan action through language reasoning. However, this inefficient paradigm fundamentally restricts agents to basic and non-fluent interactions: relying on isolated VLM reasoning for each action makes it impossible to handle tasks requiring high reactivity (e.g., FPS shooting) or dynamic adaptability (e.g., ACT combat). To handle this, we propose a paradigm shift in gameplay agent design: instead of directly controlling gameplay, VLM develops specialized execution modules tailored for tasks like shooting and combat. These modules handle real-time game interactions, elevating VLM to a high-level developer. Building upon this paradigm, we introduce GameSense, a gameplay agent framework where VLM develops task-specific game sense modules by observing task execution and leveraging vision tools and neural network training pipelines. These modules encapsulate action-feedback logic, ranging from direct action rules to neural network-based decisions. Experiments demonstrate that our framework is the first to achieve fluent gameplay in diverse genres, including ACT, FPS, and Flappy Bird, setting a new benchmark for game-playing agents.

cs.CL↗

arXiv2024

360VFI: A Dataset and Benchmark for Omnidirectional Video Frame Interpolation

Head-mounted 360° displays and portable 360° cameras have significantly progressed, providing viewers a realistic and immersive experience. However, many omnidirectional videos have low frame rates that can lead to visual fatigue, and the prevailing plane frame interpolation methodologies are unsuitable for omnidirectional video interpolation because they are designed solely for traditional videos. This paper introduces the benchmark dataset, 360VFI, for Omnidirectional Video Frame Interpolation. We present a practical implementation that introduces a distortion prior from omnidirectional video into the network to modulate distortions. Specifically, we propose a pyramid distortion-sensitive feature extractor that uses the unique characteristics of equirectangular projection (ERP) format as prior information. Moreover, we devise a decoder that uses an affine transformation to further facilitate the synthesis of intermediate frames. 360VFI is the first dataset and benchmark that explores the challenge of Omnidirectional Video Frame Interpolation. Through our benchmark analysis, we present four different distortion condition scenes in the proposed 360VFI dataset to evaluate the challenges triggered by distortion during interpolation. Besides, experimental results demonstrate that Omnidirectional Video Interpolation can be effectively improved by modeling for omnidirectional distortion.

cs.CV↗

arXiv2012

Stability Conditions and Mirror Symmetry of K3 Surfaces in Attractor Backgrounds

We study the space of stability conditions on $K3$ surfaces from the perspective of mirror symmetry. It is done in the so called attractor backgrounds (moduli) which can be far from the conventional large complex limits and are selected by the attractor mechanism for certain black holes. We find certain highly non-generic behaviors of stability walls (a key notion in the study of wall crossings) in the space of stability conditions. They correspond via mirror symmetry to some non-generic behaviors of special Lagrangians in an attractor background. The main results can be understood as a mirror correspondence in a synthesis of homological mirror conjecture and SYZ mirror conjecture.

math.AG↗

arXiv2012

SYZ Mirror Symmetry of Hitchin's Moduli Spaces Near Singular Fibers I

We study hyperkahler metrics and hyperholomorphic connections of Hitchin's moduli spaces after Gaiotto, Moore and Neitzke. Their construction via the twistor technique produces intricate wall crossing behaviors. For certain four dimensional Hitchin's moduli spaces local models and degeneration to local models near singular fibers of the Hitchin's fibration are understood.

math.DG↗

arXiv2011

Instanton Correction, Wall Crossing And Mirror Symmetry Of Hitchin's Moduli Spaces

We study two instanton correction problems of Hitchin's moduli spaces along with their wall crossing formulas. The hyperkahler metric of a Hitchin's moduli space can be put into an instanton-corrected form according to physicists Gaiotto, Moore and Neitzke. The problem boils down to the construction of a set of special coordinates which can be constructed as Fock-Goncharov coordinates associated with foliations of quadratic differentials on a Riemann surface. A wall crossing formula of Kontsevich and Soibelman arises both as a crucial consistency condition and an effective computational tool. On the other hand Gross and Siebert have succeeded in determining instanton corrections of complex structures of Calabi-Yau varieties in the context of mirror symmetry from a singular affine structure with additional data. We will show that the two instanton correction problems are equivalent in an appropriate sense via the identification of the wall crossing formulas in the metric problem with consistency conditions in the complex structure problem. This result provides examples of Calabi-Yau varieties where the instanton correction (in the sense of mirror symmetry) of metrics and complex structures can be determined.

math.AG↗