arXiv subjects

Xiaoyi Zhang

Publications and source records attributed to Xiaoyi Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Generative Video Compression with One-Dimensional Latent Representation

Recent advancements in generative video codec (GVC) typically encode video into a 2D latent grid and employ high-capacity generative decoders for reconstruction. However, this paradigm still leaves two key challenges in fully exploiting spatial-temporal redundancy: Spatially, the 2D latent grid inevitably preserves intra-frame redundancy due to its rigid structure, where adjacent patches remain highly similar, thereby necessitating a higher bitrate. Temporally, the 2D latent grid is less effective for modeling long-term correlations in a compact and semantically coherent manner, as it hinders the aggregation of common contents across frames. To address these limitations, we introduce Generative Video Compression with One-Dimensional (1D) Latent Representation (GVC1D). GVC1D encodes the video data into extreme compact 1D latent tokens conditioned on both short- and long-term contexts. Without the rigid 2D spatial correspondence, these 1D latent tokens can adaptively attend to semantic regions and naturally facilitate token reduction, thereby reducing spatial redundancy. Furthermore, the proposed 1D memory provides semantically rich long-term context while maintaining low computational cost, thereby further reducing temporal redundancy. Experimental results indicate that GVC1D attains superior compression efficiency, where it achieves bitrate reductions of 60.4\% under LPIPS and 68.8\% under DISTS on the HEVC Class B dataset, surpassing the previous video compression methods.Project: https://gvc1d.github.io/

cs.CV↗

arXiv2026

The uniqueness of the ground state and the dynamics of nonlinear Schrödinger equation with inverse square potential

In this paper, we first provide an alternative proof of the uniqueness of the ground state solution for NLS with inverse square potential and power nonlinearity $|u|^pu$ for all $0<p<\frac 4{d-2}$ in dimensions $d\ge 3$. While the uniqueness result was previously obtained by Mukherjee-Nam-Nguyen using a functional analytic approach, our method successfully adapts the classical ``shooting method'' to the case with the singular potential, accompanied by a more detailed analysis on the ground state equation. Based upon this result and a comprehensive spectral analysis, we construct the stable/unstable manifolds of the ground state standing wave solutions and classify solutions on the mass-energy level surface of the ground state in dimensions $d=3, 4, 5$.

math.AP↗

arXiv2026

From Off-Policy to On-Policy: Enhancing GUI Agents via Bi-level Expert-to-Policy Assimilation

Vision-language models are increasingly deployed as computer-use agents (CUAs) that operate desktops and browsers. Top-performing CUAs are framework-based systems that decompose planning and execution, while end-to-end screenshot-to-action policies are easier to deploy but lag behind on benchmarks such as OSWorld-Verified. GUI datasets like OSWorld pose two bottlenecks: they expose only a few hundred interactive, verifiable tasks and environments, and expert trajectories must be gathered by interacting with these environments, making such data hard to scale. We therefore ask how reinforcement learning from verifiable rewards (RLVR) can best exploit a small pool of exist expert trajectories to train end-to-end policies. Naively mixing these off-policy traces into on-policy RLVR is brittle: even after format conversion, expert trajectories exhibit structural mismatch and distribution shift from the learner. We propose BEPA (Bi-Level Expert-to-Policy Assimilation), which turns static expert traces into policy-aligned guidance via self-rolled reachable trajectories under the base policy (LEVEL-1) and a per-task, dynamically updated cache used in RLVR (LEVEL-2). On OSWorld-Verified, BEPA improves UITARS1.5-7B success from 22.87% to 32.13% and raises a held-out split from 5.74% to 10.30%, with consistent gains on MMBench-GUI and Online-Mind2Web. Our code and data are available at: https://github.com/LEON-gittech/Verl_GUI.git

cs.AI↗

arXiv2026

Photoinduced metastable cation disorder in metal halide double perovskites

Lead-free perovskites have emerged as environmentally benign alternatives to lead-halide counterparts for optoelectronics. Among them, the double perovskite Cs2AgInCl6 family exhibits remarkable white-light emission with proper composition engineering, enabled by strong electron-phonon coupling and the formation of self-trapped excitons (STEs). Despite these advantages, the fundamental photo- and structural dynamics governing their excited-state behavior remain poorly understood. Here, we report a long-lived metastable phase in the Cs2AgInCl6 double perovskite family and unravel this process and the concomitant electronic and structural evolution using a suite of tools including transient optical spectroscopy, time-resolved X-ray diffraction (TR-XRD) and X-ray absorption (TR-XAS). We show that the photoinduced, transient metastable phase is associated with B-site (Ag-In) disorder, which induces a dramatically reduced optical bandgap. Supported by TR-XRD and first-principles calculations, the Ag-In disorder drives the formation of Ag-rich and In-rich domains with millisecond lifetimes, with lifetimes increasing at lower temperatures. TR-XAS further reveals that photogenerated STEs oxidize Ag+ to Ag2+, facilitating this highly temporally asymmetric order-disorder transition. Our findings demonstrate a new mechanism, mediated by hole-localized STE formation, that enables prolongation of transient light-induced states to the multi-millisecond regime in double perovskites, opening possibilities to harvesting the functional properties of metastable phases of these materials.

cond-mat.mtrl-sci↗

arXiv2025

Deep Video Discovery: Agentic Search with Tool Use for Long-form Video Understanding

Long-form video understanding presents significant challenges due to extensive temporal-spatial complexity and the difficulty of question answering under such extended contexts. While Large Language Models (LLMs) have demonstrated considerable advancements in video analysis capabilities and long context handling, they continue to exhibit limitations when processing information-dense hour-long videos. To overcome such limitations, we propose the Deep Video Discovery (DVD) agent to leverage an agentic search strategy over segmented video clips. Unlike previous video agents that rely on predefined workflows applied uniformly across different queries, our approach emphasizes the autonomous and adaptive nature of agents. By providing a set of search-centric tools on multi-granular video database, our DVD agent leverages the advanced reasoning capability of LLM to plan on its current observation state, strategically selects tools to orchestrate adaptive workflow for different queries in light of the gathered information. We perform comprehensive evaluation on multiple long video understanding benchmarks that demonstrates our advantage. Our DVD agent achieves state-of-the-art performance on the challenging LVBench dataset, reaching an accuracy of 74.2%, which substantially surpasses all prior works, and further improves to 76.0% with transcripts. The code has been released at https://github.com/microsoft/DeepVideoDiscovery.

cs.CV↗

arXiv2025

UI-Evol: Automatic Knowledge Evolving for Computer Use Agents

External knowledge has played a crucial role in the recent development of computer use agents. We identify a critical knowledge-execution gap: retrieved knowledge often fails to translate into effective real-world task execution. Our analysis shows even 90% correct knowledge yields only 41% execution success rate. To bridge this gap, we propose UI-Evol, a plug-and-play module for autonomous GUI knowledge evolution. UI-Evol consists of two stages: a Retrace Stage that extracts faithful objective action sequences from actual agent-environment interactions, and a Critique Stage that refines existing knowledge by comparing these sequences against external references. We conduct comprehensive experiments on the OSWorld benchmark with the state-of-the-art Agent S2. Our results demonstrate that UI-Evol not only significantly boosts task performance but also addresses a previously overlooked issue of high behavioral standard deviation in computer use agents, leading to superior performance on computer use tasks and substantially improved agent reliability.

cs.HC↗

arXiv2025

Generative Latent Video Compression

Perceptual optimization is widely recognized as essential for neural compression, yet balancing the rate-distortion-perception tradeoff remains challenging. This difficulty is especially pronounced in video compression, where frame-wise quality fluctuations often cause perceptually optimized neural video codecs to suffer from flickering artifacts. In this paper, inspired by the success of latent generative models, we present Generative Latent Video Compression (GLVC), an effective framework for perceptual video compression. GLVC employs a pretrained continuous tokenizer to project video frames into a perceptually aligned latent space, thereby offloading perceptual constraints from the rate-distortion optimization. We redesign the codec architecture explicitly for the latent domain, drawing on extensive insights from prior neural video codecs, and further equip it with innovations such as unified intra/inter coding and a recurrent memory mechanism. Experimental results across multiple benchmarks show that GLVC achieves state-of-the-art performance in terms of DISTS and LPIPS metrics. Notably, our user study confirms GLVC rivals the latest neural video codecs at nearly half their rate while maintaining stable temporal coherence, marking a step toward practical perceptual video compression.

eess.IV↗

arXiv2025

Attack for Defense: Adversarial Agents for Point Prompt Optimization Empowering Segment Anything Model

Prompt quality plays a critical role in the performance of the Segment Anything Model (SAM), yet existing approaches often rely on heuristic or manually crafted prompts, limiting scalability and generalization. In this paper, we propose Point Prompt Defender, an adversarial reinforcement learning framework that adopts an attack-for-defense paradigm to automatically optimize point prompts. We construct a task-agnostic point prompt environment by representing image patches as nodes in a dual-space graph, where edges encode both physical and semantic distances. Within this environment, an attacker agent learns to activate a subset of prompts that maximally degrade SAM's segmentation performance, while a defender agent learns to suppress these disruptive prompts and restore accuracy. Both agents are trained using Deep Q-Networks with a reward signal based on segmentation quality variation. During inference, only the defender is deployed to refine arbitrary coarse prompt sets, enabling enhanced SAM segmentation performance across diverse tasks without retraining. Extensive experiments show that Point Prompt Defender effectively improves SAM's robustness and generalization, establishing a flexible, interpretable, and plug-and-play framework for prompt-based segmentation.

cs.CV↗

arXiv2025

UI-E2I-Synth: Advancing GUI Grounding with Large-Scale Instruction Synthesis

Recent advancements in Large Vision-Language Models are accelerating the development of Graphical User Interface (GUI) agents that utilize human-like vision perception capabilities to enhance productivity on digital devices. Compared to approaches predicated on GUI metadata, which are platform-dependent and vulnerable to implementation variations, vision-based approaches offer broader applicability. In this vision-based paradigm, the GUI instruction grounding, which maps user instruction to the location of corresponding element on the given screenshot, remains a critical challenge, particularly due to limited public training dataset and resource-intensive manual instruction data annotation. In this paper, we delve into unexplored challenges in this task including element-to-screen ratio, unbalanced element type, and implicit instruction. To address these challenges, we introduce a large-scale data synthesis pipeline UI-E2I-Synth for generating varying complex instruction datasets using GPT-4o instead of human annotators. Furthermore, we propose a new GUI instruction grounding benchmark UI-I2E-Bench, which is designed to address the limitations of existing benchmarks by incorporating diverse annotation aspects. Our model, trained on the synthesized data, achieves superior performance in GUI instruction grounding, demonstrating the advancements of proposed data synthesis pipeline. The proposed benchmark, accompanied by extensive analyses, provides practical insights for future research in GUI grounding. We will release corresponding artifacts at https://microsoft.github.io/FIVE-UI-Evol/ .

cs.HC↗

arXiv2025

Enhanced oil recovery in reservoirs via diffusion-driven $\text{CO}_{2}$ flooding: Experimental insights and material balance modeling

$\text{CO}_{2}$ flooding is central to carbon utilization technologies, yet conventional waterflooding models fail to capture the complex interactions between CO$_2$ and formation fluids. In this study, one- and two-dimensional nuclear magnetic resonance experiments reveal that $\text{CO}_{2}$ markedly enhances crude oil mobility during miscible displacement via multiple synergistic mechanisms, yielding a recovery factor of $60.97\%$, which surpasses that of immiscible displacement (maximum $57.53\%$). Guided by these findings, we propose a convection-diffusion model that incorporates the diffusion coefficient ($D$) and porosity ($ϕ$) as key parameters. This model captures the spatiotemporal evolution of the $\text{CO}_{2}$ front and addresses a key limitation of conventional formulations-the omission of diffusion effects. It improves predictions of gas breakthrough time and enables optimized injection design for low-permeability reservoirs. Extending classical material balance theory, we develop an enhanced $\text{CO}_{2}$ flooding equation that integrates critical transport phenomena. This formulation incorporates $\text{CO}_{2}$ diffusion, oil phase expansion, reservoir adsorption, and gas compressibility to describe the dynamic transport and mass compensation of injected $\text{CO}_{2}$. Validation through experimental and numerical data confirms the model's robustness and applicability under low-permeability conditions. The proposed framework overcomes limitations of physical experiments under extreme environments and offers theoretical insight into oil recovery enhancement and $\text{CO}_{2}$ injection strategy optimization.

physics.flu-dyn↗

arXiv2024

Ultrafast symmetry control in photoexcited quantum dots

Symmetry control is essential for realizing unconventional properties, such as ferroelectricity, nonlinear optical responses, and complex topological order, thus it holds promise for the design of emerging quantum and photonic systems. Nevertheless, fast and reversible control of symmetry in materials remains a challenge, especially for nanoscale systems. Here, we unveil reversible symmetry changes in colloidal lead chalcogenide quantum dots on picosecond timescales. Using a combination of ultrafast electron diffraction and total X-ray scattering, in conjunction with atomic-scale structural modeling and first-principles calculations, we reveal that symmetry-broken lead sulfide quantum dots restore to a centrosymmetric phase upon photoexcitation. The symmetry restoration is driven by photoexcited electronic carriers, which suppress lead off-centering for about 100 ps. Furthermore, the change in symmetry is closely correlated with the electronic properties as shown by transient optical measurements. Overall, this study elucidates reversible symmetry changes in colloidal quantum dots, and more broadly defines a new methodology to optically control symmetry in nanoscale systems on ultrafast timescales.

cond-mat.mes-hall↗

arXiv2024

Text Grouping Adapter: Adapting Pre-trained Text Detector for Layout Analysis

Significant progress has been made in scene text detection models since the rise of deep learning, but scene text layout analysis, which aims to group detected text instances as paragraphs, has not kept pace. Previous works either treated text detection and grouping using separate models, or train a model from scratch while using a unified one. All of them have not yet made full use of the already well-trained text detectors and easily obtainable detection datasets. In this paper, we present Text Grouping Adapter (TGA), a module that can enable the utilization of various pre-trained text detectors to learn layout analysis, allowing us to adopt a well-trained text detector right off the shelf or just fine-tune it efficiently. Designed to be compatible with various text detector architectures, TGA takes detected text regions and image features as universal inputs to assemble text instance features. To capture broader contextual information for layout analysis, we propose to predict text group masks from text instance features by one-to-many assignment. Our comprehensive experiments demonstrate that, even with frozen pre-trained models, incorporating our TGA into various pre-trained text detectors and text spotters can achieve superior layout analysis performance, simultaneously inheriting generalized text detection ability from pre-training. In the case of full parameter fine-tuning, we can further improve layout analysis performance.

cs.CV↗

arXiv2024

Talaria: Interactively Optimizing Machine Learning Models for Efficient Inference

On-device machine learning (ML) moves computation from the cloud to personal devices, protecting user privacy and enabling intelligent user experiences. However, fitting models on devices with limited resources presents a major technical challenge: practitioners need to optimize models and balance hardware metrics such as model size, latency, and power. To help practitioners create efficient ML models, we designed and developed Talaria: a model visualization and optimization system. Talaria enables practitioners to compile models to hardware, interactively visualize model statistics, and simulate optimizations to test the impact on inference metrics. Since its internal deployment two years ago, we have evaluated Talaria using three methodologies: (1) a log analysis highlighting its growth of 800+ practitioners submitting 3,600+ models; (2) a usability survey with 26 users assessing the utility of 20 Talaria features; and (3) a qualitative interview with the 7 most active users about their experience using Talaria.

cs.HC↗

arXiv2023

Responsible Task Automation: Empowering Large Language Models as Responsible Task Automators

The recent success of Large Language Models (LLMs) signifies an impressive stride towards artificial general intelligence. They have shown a promising prospect in automatically completing tasks upon user instructions, functioning as brain-like coordinators. The associated risks will be revealed as we delegate an increasing number of tasks to machines for automated completion. A big question emerges: how can we make machines behave responsibly when helping humans automate tasks as personal copilots? In this paper, we explore this question in depth from the perspectives of feasibility, completeness and security. In specific, we present Responsible Task Automation (ResponsibleTA) as a fundamental framework to facilitate responsible collaboration between LLM-based coordinators and executors for task automation with three empowered capabilities: 1) predicting the feasibility of the commands for executors; 2) verifying the completeness of executors; 3) enhancing the security (e.g., the protection of users' privacy). We further propose and compare two paradigms for implementing the first two capabilities. One is to leverage the generic knowledge of LLMs themselves via prompt engineering while the other is to adopt domain-specific learnable models. Moreover, we introduce a local memory mechanism for achieving the third capability. We evaluate our proposed ResponsibleTA on UI task automation and hope it could bring more attentions to ensuring LLMs more responsible in diverse scenarios.

cs.AI↗

arXiv2023

GO-FEAP: Global Optimal UAV Planner Using Frontier-Omission-Aware Exploration and Altitude-Stratified Planning

Autonomous exploration is a fundamental problem for various applications of unmanned aerial vehicles(UAVs). Existing methods, however, are demonstrated to static local optima and two-dimensional exploration. To address these challenges, this paper introduces GO-FEAP (Global Optimal UAV Planner Using Frontier-Omission-Aware Exploration and Altitude-Stratified Planning), aiming to achieve efficient and complete three-dimensional exploration. Frontier-Omission-Aware Exploration module presented in this work takes into account multiple pivotal factors, encompassing frontier distance, nearby frontier count, frontier duration, and frontier categorization, for a comprehensive assessment of frontier importance. Furthermore, to tackle scenarios with substantial vertical variations, we introduce the Altitude-Stratified Planning strategy, which stratifies the three-dimensional space based on altitude, conducting global-local planning for each stratum. The objective of global planning is to identify the most optimal frontier for exploration, followed by viewpoint selection and local path optimization based on frontier type, ultimately generating dynamically feasible three-dimensional spatial exploration trajectories. We present extensive benchmark and real-world tests, in which our method completes the exploration tasks with unprecedented completeness compared to state-of-the-art approaches.

cs.RO↗

arXiv2023

Towards Automated Accessibility Report Generation for Mobile Apps

Many apps have basic accessibility issues, like missing labels or low contrast. Automated tools can help app developers catch basic issues, but can be laborious or require writing dedicated tests. We propose a system, motivated by a collaborative process with accessibility stakeholders at a large technology company, to generate whole app accessibility reports by combining varied data collection methods (e.g., app crawling, manual recording) with an existing accessibility scanner. Many such scanners are based on single-screen scanning, and a key problem in whole app accessibility reporting is to effectively de-duplicate and summarize issues collected across an app. To this end, we developed a screen grouping model with 96.9% accuracy (88.8% F1-score) and UI element matching heuristics with 97% accuracy (98.2% F1-score). We combine these technologies in a system to report and summarize unique issues across an app, and enable a unique pixel-based ignore feature to help engineers and testers better manage reported issues across their app's lifetime. We conducted a qualitative evaluation with 18 accessibility-focused engineers and testers which showed this system can enhance their existing accessibility testing toolkit and address key limitations in current accessibility scanning tools.

cs.HC↗

arXiv2023

Reinforced UI Instruction Grounding: Towards a Generic UI Task Automation API

Recent popularity of Large Language Models (LLMs) has opened countless possibilities in automating numerous AI tasks by connecting LLMs to various domain-specific models or APIs, where LLMs serve as dispatchers while domain-specific models or APIs are action executors. Despite the vast numbers of domain-specific models/APIs, they still struggle to comprehensively cover super diverse automation demands in the interaction between human and User Interfaces (UIs). In this work, we build a multimodal model to ground natural language instructions in given UI screenshots as a generic UI task automation executor. This metadata-free grounding model, consisting of a visual encoder and a language decoder, is first pretrained on well studied document understanding tasks and then learns to decode spatial information from UI screenshots in a promptable way. To facilitate the exploitation of image-to-text pretrained knowledge, we follow the pixel-to-sequence paradigm to predict geometric coordinates in a sequence of tokens using a language decoder. We further propose an innovative Reinforcement Learning (RL) based algorithm to supervise the tokens in such sequence jointly with visually semantic metrics, which effectively strengthens the spatial decoding capability of the pixel-to-sequence paradigm. Extensive experiments demonstrate our proposed reinforced UI instruction grounding model outperforms the state-of-the-art methods by a clear margin and shows the potential as a generic UI task automation API.

cs.CV↗

arXiv2023

Intestinal Microecology in Pediatric Surgery-Related Gastrointestinal Diseases Current Insights and Future Perspectives

Intestinal microecology is established from birth and is constantly changing until homeostasis is reached. Intestinal microecology is involved in the immune inflammatory response of the intestine and regulates the intestinal barrier function. The imbalance of intestinal microecology is closely related to the occurrence and development of digestive system diseases. In some gastrointestinal diseases related to pediatric surgery, intestinal microecology and its metabolites undergo a series of changes, which can provide a certain basis for the diagnosis of diseases. The continuous development of microecological agents and fecal microbiota transplantation technology has provided a new means for its clinical treatment. We review the relationship between pathogenesis, diagnosis and treatment of pediatric surgery-related gastrointestinal diseases and intestinal microecology, in order to provide new ideas and methods for clinical diagnosis, treatment and research.

q-bio.TO↗