arXiv subjects

Yuqi Zhou

Publications and source records attributed to Yuqi Zhou.

At least 19 recordsLinked to original sources

arXiv2026

TurnSight: Turn-Level Hindsight Self-Distillation for Tool-Integrated Reasoning

Tool-Integrated Reasoning (TIR) enables LLMs to solve complex tasks through iterative tool interactions. However, existing reinforcement learning methods often rely on trajectory-level supervision, limiting fine-grained credit assignment in long-horizon TIR scenarios. On-policy self-distillation offers denser signals through teacher branches with privileged context, but existing approaches typically derive such context from ground-truth answers or retrieved skills, which may not reflect the states actually visited by the agent. Moreover, token-level supervision fails to capture the turn-level structure of tool interactions. To address this, we propose TurnSight, a turn-level hindsight self-distillation framework that derives supervision directly from execution-conditioned hindsight. It then constructs multiple hindsight views with different lookahead horizons and selects reliable supervision through cross-horizon directional agreement. Finally, the selected hindsight signal is normalized across sibling rollouts and used to adaptively modulate RL advantages while preserving their original optimization direction. Extensive experiments on three benchmarks demonstrate the effectiveness of TurnSight. Our codes are available at https://github.com/quchangle1/TurnSight.

cs.CL↗

arXiv2026

DiffLens: A Visualization System to Explore Local Differences in Graph Sampling

Graph sampling techniques have been widely used to simplify network computation and visualization, which also results in inevitable differences between the sampled networks and the original networks in terms of nodes, edges and structures. Investigating such differences can inform graph sampling technique users of the pros and cons of different techniques and select the appropriate one, and can also help graph sampling developers evaluate their own technique. However, there are still no systematic ways to achieve such a goal. This paper fills this research gap by first proposing systematic and generic quantitative measures to quantify three categories of graph differences (i.e., neighbor-based, path-based, and structure-based). Built upon this, we further propose DiffLens, a novel visualization system to help graph sampling developers and users intuitively explore local differences at different regions of their interest within a sampled graph, where three new lens-based visual designs are presented to display the neighbor-based, path-based, and structure-based differences respectively. We conducted two case studies and a user study using real-world network datasets to evaluate DiffLens. The results confirmed its effectiveness and usability in helping users explore local differences and compare different graph sampling strategies.

cs.HC↗

arXiv2026

Temporal Dynamical Quantum Phase Transition in Dicke Model with Trapped Ions

Temporal non-analyticities in the rate function of the Loschmidt echo manifests a class of dynamical quantum phase transitions (DQPTs) that has emerged as a powerful framework for understanding far-from-equilibrium many-body dynamics. While such DQPT has been extensively studied theoretically in spin-boson systems such as the Dicke model, their experimental observation remains elusive. In particular, the dynamics of DQPT in asymmetric spin subspaces and under the influence of spin dissipation are largely unexplored. Here, we report an experimental study of temporal DQPT in a generalized Dicke model using a trapped-ion quantum simulator. By coupling a linear chain of $\rm{^{40}Ca^{+}}$ ions to a collective center-of-mass motional mode, we probe the quench dynamics starting from both symmetric and asymmetric initial states. We extract the rate function and identify temporal turn-around points that are in quantitative agreement with theoretical predictions. Additionally, we investigate the impact of spin dissipation on these dynamics. Our results establish an experimental platform for probing complex many-body out-of-equilibrium phenomena and advance the development of hybrid oscillator-spin quantum simulators.

quant-ph↗

arXiv2026

STEPIC: High-Speed Imaging via Spatio-Temporal Encoding in Photonic Integrated Circuits

High-speed imaging of cells in flow is essential for probing cellular heterogeneity in large populations. Existing imaging approaches based on single-pixel detection and spatio-temporal encoding provide exceptional speed, but typically rely on bulky free-space optics, long dispersive elements, and are prone to alignment instabilities. Here, we introduce STEPIC Microscopy, the first fully integrated on-chip system for high-speed imaging via spatio-temporal encoding in photonic integrated circuits. Our platform leverages waveguides, splitters, fiber delay-lines, and 3D optical remappers to encode spatial information into the temporal domain, enabling robust image reconstruction of cells flowing through microchannels. The monolithic architecture provides a compact and robust platform for high-throughput bioimaging, enabling scalable and practical implementations of ultrafast imaging systems.

physics.optics↗

arXiv2026

DR-Venus: Towards Frontier Edge-Scale Deep Research Agents with Only 10K Open Data

Edge-scale deep research agents based on small language models are attractive for real-world deployment due to their advantages in cost, latency, and privacy. In this work, we study how to train a strong small deep research agent under limited open-data by improving both data quality and data utilization. We present DR-Venus, a frontier 4B deep research agent for edge-scale deployment, built entirely on open data. Our training recipe consists of two stages. In the first stage, we use agentic supervised fine-tuning (SFT) to establish basic agentic capability, combining strict data cleaning with resampling of long-horizon trajectories to improve data quality and utilization. In the second stage, we apply agentic reinforcement learning (RL) to further improve execution reliability on long-horizon deep research tasks. To make RL effective for small agents in this setting, we build on IGPO and design turn-level rewards based on information gain and format-aware regularization, thereby enhancing supervision density and turn-level credit assignment. Built entirely on roughly 10K open-data, DR-Venus-4B significantly outperforms prior agentic models under 9B parameters on multiple deep research benchmarks, while also narrowing the gap to much larger 30B-class systems. Our further analysis shows that 4B agents already possess surprisingly strong performance potential, highlighting both the deployment promise of small models and the value of test-time scaling in this setting. We release our models, code, and key recipes to support reproducible research on edge-scale deep research agents.

cs.LG↗

arXiv2026

Learning to Retrieve from Agent Trajectories

Information retrieval (IR) systems have traditionally been designed and trained for human users, with learning-to-rank methods relying heavily on large-scale human interaction logs such as clicks and dwell time. With the rapid emergence of large language model (LLM) powered search agents, however, retrieval is increasingly consumed by agents rather than human beings, and is embedded as a core component within multi-turn reasoning and action loops. In this setting, retrieval models trained under human-centric assumptions exhibit a fundamental mismatch with the way agents issue queries and consume results. In this work, we argue that retrieval models for agentic search should be trained directly from agent interaction data. We introduce learning to retrieve from agent trajectories as a new training paradigm, where supervision is derived from multi-step agent interactions. Through a systematic analysis of search agent trajectories, we identify key behavioral signals that reveal document utility, including browsing actions, unbrowsed rejections, and post-browse reasoning traces. Guided by these insights, we propose LRAT, a simple yet effective framework that mines high-quality retrieval supervision from agent trajectories and incorporates relevance intensity through weighted optimization. Extensive experiments on both in-domain and out-of-domain deep research benchmarks demonstrate that retrievers trained with LRAT consistently improve evidence recall, end-to-end task success, and execution efficiency across diverse agent architectures and scales. Our results highlight agent trajectories as a practical and scalable supervision source, pointing to a promising direction for retrieval in the era of agentic search.

cs.IR↗

arXiv2026

UI-Venus-1.5 Technical Report

GUI agents have emerged as a powerful paradigm for automating interactions in digital environments, yet achieving both broad generality and consistently strong task performance remains challenging. In this report, we present UI-Venus-1.5, a unified, end-to-end GUI Agent designed for robust real-world applications. The proposed model family comprises two dense variants (2B and 8B) and one mixture-of-experts variant (30B-A3B) to meet various downstream application scenarios. Compared to our previous version, UI-Venus-1.5 introduces three key technical advances: (1) a comprehensive Mid-Training stage leveraging 10 billion tokens across 30+ datasets to establish foundational GUI semantics; (2) Online Reinforcement Learning with full-trajectory rollouts, aligning training objectives with long-horizon, dynamic navigation in large-scale environments; and (3) a single unified GUI Agent constructed via Model Merging, which synthesizes domain-specific models (grounding, web, and mobile) into one cohesive checkpoint. Extensive evaluations demonstrate that UI-Venus-1.5 establishes new state-of-the-art performance on benchmarks such as ScreenSpot-Pro (69.6%), VenusBench-GD (75.0%), and AndroidWorld (77.6%), significantly outperforming previous strong baselines. In addition, UI-Venus-1.5 demonstrates robust navigation capabilities across a variety of Chinese mobile apps, effectively executing user instructions in real-world scenarios. Code: https://github.com/inclusionAI/UI-Venus; Model: https://huggingface.co/collections/inclusionAI/ui-venus

cs.CV↗

arXiv2026

VenusBench-Mobile: A Challenging and User-Centric Benchmark for Mobile GUI Agents with Capability Diagnostics

Existing online benchmarks for mobile GUI agents remain largely app-centric and task-homogeneous, failing to reflect the diversity and instability of real-world mobile usage. To this end, we introduce VenusBench-Mobile, a challenging online benchmark for evaluating general-purpose mobile GUI agents under realistic, user-centric conditions. VenusBench-Mobile builds two core evaluation pillars: defining what to evaluate via user-intent-driven task design that reflects real mobile usage, and how to evaluate through a capability-oriented annotation scheme for fine-grained agent behavior analysis. Extensive evaluation of state-of-the-art mobile GUI agents reveals large performance gaps relative to prior benchmarks, indicating that VenusBench-Mobile poses substantially more challenging and realistic tasks and that current agents remain far from reliable real-world deployment. Diagnostic analysis further shows that failures are dominated by deficiencies in perception and memory, which are largely obscured by coarse-grained evaluations. Moreover, even the strongest agents exhibit near-zero success under environment variations, highlighting their brittleness in realistic settings. Based on these insights, we believe VenusBench-Mobile provides an important stepping stone toward robust real-world deployment of mobile GUI agents. Code and data are available at https://github.com/inclusionAI/UI-Venus/tree/VenusBench-Mobile.

cs.HC↗

arXiv2025

Inferring Causal Relationships to Improve Caching for Clients with Correlated Requests: Applications to VR

Efficient edge caching reduces latency and alleviates backhaul congestion in modern networks. Traditional caching policies, such as Least Recently Used (LRU) and Least Frequently Used (LFU), perform well under specific request patterns. LRU excels in workloads with strong temporal locality, while LFU is effective when content popularity remains static. However, real-world client requests often exhibit correlations due to shared contexts and coordinated activities. This is particularly evident in Virtual Reality (VR) environments, where groups of clients navigate shared virtual spaces, leading to correlated content requests. In this paper, we introduce the \textit{grouped client request model}, a generalization of the Independent Reference Model that explicitly captures different types of request correlations. Our theoretical analysis of LRU under this model reveals that the optimal causal caching policy depends on cache size: LFU is optimal for small to moderate caches, while LRU outperforms it for larger caches. To address the limitations of existing policies, we propose Least Following and Recently Used (LFRU), a novel online caching policy that dynamically infers and adapts to causal relationships in client requests to optimize evictions. LFRU prioritizes objects likely to be requested based on inferred dependencies, achieving near-optimal performance compared to the offline optimal Belady policy in structured correlation settings. We develop VR based datasets to evaluate caching policies under realistic correlated requests. Our results show that LFRU consistently performs at least as well as LRU and LFU, outperforming LRU by up to 2.9x and LFU by up to1.9x in certain settings.

cs.NI↗

arXiv2025

PolarStore: High-Performance Data Compression for Large-Scale Cloud-Native Databases

In recent years, resource elasticity and cost optimization have become essential for RDBMSs. While cloud-native RDBMSs provide elastic computing resources via disaggregated computing and storage, storage costs remain a critical user concern. Consequently, data compression emerges as an effective strategy to reduce storage costs. However, existing compression approaches in RDBMSs present a stark trade-off: software-based approaches incur significant performance overheads, while hardware-based alternatives lack the flexibility required for diverse database workloads. In this paper, we present PolarStore, a compressed shared storage system for cloud-native RDBMSs. PolarStore employs a dual-layer compression mechanism that combines in-storage compression in PolarCSD hardware with lightweight compression in software. This design leverages the strengths of both approaches. PolarStore also incorporates database-oriented optimizations to maintain high performance on critical I/O paths. Drawing from large-scale deployment experiences, we also introduce hardware improvements for PolarCSD to ensure host-level stability and propose a compression-aware scheduling scheme to improve cluster-level space efficiency. PolarStore is currently deployed on thousands of storage servers within PolarDB, managing over 100 PB of data. It achieves a compression ratio of 3.55 and reduces storage costs by approximately 60%. Remarkably, these savings are achieved while maintaining performance comparable to uncompressed clusters.

cs.DC↗

arXiv2025

Beluga: A CXL-Based Memory Architecture for Scalable and Efficient LLM KVCache Management

The rapid increase in LLM model sizes and the growing demand for long-context inference have made memory a critical bottleneck in GPU-accelerated serving systems. Although high-bandwidth memory (HBM) on GPUs offers fast access, its limited capacity necessitates reliance on host memory (CPU DRAM) to support larger working sets such as the KVCache. However, the maximum DRAM capacity is constrained by the limited number of memory channels per CPU socket. To overcome this limitation, current systems often adopt RDMA-based disaggregated memory pools, which introduce significant challenges including high access latency, complex communication protocols, and synchronization overhead. Fortunately, the emerging CXL technology introduces new opportunities in KVCache design. In this paper, we propose Beluga, a novel memory architecture that enables GPUs and CPUs to access a shared, large-scale memory pool through CXL switches. By supporting native load/store access semantics over the CXL fabric, our design delivers near-local memory latency, while reducing programming complexity and minimizing synchronization overhead. We conduct a systematic characterization of a commercial CXL switch-based memory pool and propose a set of design guidelines. Based on Beluga, we design and implement Beluga-KVCache, a system tailored for managing the large-scale KVCache in LLM inference. Beluga-KVCache achieves an 89.6% reduction in Time-To-First-Token (TTFT) and 7.35x throughput improvement in the vLLM inference engine compared to RDMA-based solutions. To the best of our knowledge, Beluga is the first system that enables GPUs to directly access large-scale memory pools through CXL switches, marking a significant step toward low-latency, shared access to vast memory resources by GPUs.

cs.DC↗

arXiv2025

Joint Deblurring and 3D Reconstruction for Macrophotography

Macro lens has the advantages of high resolution and large magnification, and 3D modeling of small and detailed objects can provide richer information. However, defocus blur in macrophotography is a long-standing problem that heavily hinders the clear imaging of the captured objects and high-quality 3D reconstruction of them. Traditional image deblurring methods require a large number of images and annotations, and there is currently no multi-view 3D reconstruction method for macrophotography. In this work, we propose a joint deblurring and 3D reconstruction method for macrophotography. Starting from multi-view blurry images captured, we jointly optimize the clear 3D model of the object and the defocus blur kernel of each pixel. The entire framework adopts a differentiable rendering method to self-supervise the optimization of the 3D model and the defocus blur kernel. Extensive experiments show that from a small number of multi-view images, our proposed method can not only achieve high-quality image deblurring but also recover high-fidelity 3D appearance.

cs.CV↗

arXiv2025

Alternative Graph Neural Networks: Synergizing GEV Models and Deep Learning for Travel Mode Choice Modeling

Generalized extreme value models capture dependence among choice alternatives in discrete choice modeling, but require this dependence to be predefined, symmetric, and shared uniformly across individuals. Recent efforts to synergize discrete choice models with deep neural networks have improved predictive performance but still cannot explicitly represent alternative dependence within neural architectures. To address these gaps, we introduce the alternative graph -- a graph in which nodes represent choice alternatives and edges encode their dependence -- and propose Alternative Graph Neural Networks (Alt-GNNs), a family of GNN-based discrete choice models that embed alternative dependence within a unified framework. Theoretically, Alt-GNNs incorporate multinomial logit, nested logit, and ASU-DNN as special cases and enable innovative model designs, including Nested Alt-GNN, Complete Alt-GNN, and Attention Alt-GNN. Alt-GNNs are consistent with random utility maximization theory, enforce behavioral constraints through alternative graphs, and offer a novel graph-based interpretation of utility functions. Empirically, on two travel mode choice datasets from London and Chicago, Alt-GNNs significantly improve predictive performance over all benchmark models in mode choice modeling because of their flexible alternative graph design and vast hyperparameter space. Even the simplest Alt-GNN variant -- Nested Alt-GNN -- generalizes the nested logit model while preserving its unique two-layer substitution properties, enabling graph-based behavioral constraints over otherwise unconstrained behavioral patterns from deep neural networks.

stat.ML↗

arXiv2025

Graph neural networks for residential location choice: connection to classical logit models

Researchers have adopted deep learning for classical discrete choice analysis as it can capture complex feature relationships and achieve higher predictive performance. However, the existing deep learning approaches cannot explicitly capture the relationship among choice alternatives, which has been a long-lasting focus in classical discrete choice models. To address the gap, this paper introduces Graph Neural Network (GNN) as a novel framework to analyze residential location choice. The GNN-based discrete choice models (GNN-DCMs) offer a structured approach for neural networks to capture dependence among spatial alternatives, while maintaining clear connections to classical random utility theory. Theoretically, we demonstrate that the GNN-DCMs incorporate the nested logit (NL) model and the spatially correlated logit (SCL) model as two specific cases, yielding novel algorithmic interpretation through message passing among alternatives' utilities. Empirically, the GNN-DCMs outperform benchmark MNL, SCL, and feedforward neural networks in predicting residential location choices among Chicago's 77 community areas. Regarding model interpretation, the GNN-DCMs can capture individual heterogeneity and exhibit spatially-aware substitution patterns. Overall, these results highlight the potential of GNN-DCMs as a unified and expressive framework for synergizing discrete choice modeling and deep learning in the complex spatial choice contexts.

stat.ML↗

arXiv2025

GUI-G1: Understanding R1-Zero-Like Training for Visual Grounding in GUI Agents

Recent Graphical User Interface (GUI) agents replicate the R1-Zero paradigm, coupling online Reinforcement Learning (RL) with explicit chain-of-thought reasoning prior to object grounding and thereby achieving substantial performance gains. In this paper, we first conduct extensive analysis experiments of three key components of that training pipeline: input design, output evaluation, and policy update-each revealing distinct challenges arising from blindly applying general-purpose RL without adapting to GUI grounding tasks. Input design: Current templates encourage the model to generate chain-of-thought reasoning, but longer chains unexpectedly lead to worse grounding performance. Output evaluation: Reward functions based on hit signals or box area allow models to exploit box size, leading to reward hacking and poor localization quality. Policy update: Online RL tends to overfit easy examples due to biases in length and sample difficulty, leading to under-optimization on harder cases. To address these issues, we propose three targeted solutions. First, we adopt a Fast Thinking Template that encourages direct answer generation, reducing excessive reasoning during training. Second, we incorporate a box size constraint into the reward function to mitigate reward hacking. Third, we revise the RL objective by adjusting length normalization and adding a difficulty-aware scaling factor, enabling better optimization on hard samples. Our GUI-G1-3B, trained on 17K public samples with Qwen2.5-VL-3B-Instruct, achieves 90.3% accuracy on ScreenSpot and 37.1% on ScreenSpot-Pro. This surpasses all prior models of similar size and even outperforms the larger UI-TARS-7B, establishing a new state-of-the-art in GUI agent grounding. The project repository is available at https://github.com/Yuqi-Zhou/GUI-G1.

cs.CL↗

arXiv2025

Controlling Avatar Diffusion with Learnable Gaussian Embedding

Recent advances in diffusion models have made significant progress in digital human generation. However, most existing models still struggle to maintain 3D consistency, temporal coherence, and motion accuracy. A key reason for these shortcomings is the limited representation ability of commonly used control signals(e.g., landmarks, depth maps, etc.). In addition, the lack of diversity in identity and pose variations in public datasets further hinders progress in this area. In this paper, we analyze the shortcomings of current control signals and introduce a novel control signal representation that is optimizable, dense, expressive, and 3D consistent. Our method embeds a learnable neural Gaussian onto a parametric head surface, which greatly enhances the consistency and expressiveness of diffusion-based head models. Regarding the dataset, we synthesize a large-scale dataset with multiple poses and identities. In addition, we use real/synthetic labels to effectively distinguish real and synthetic data, minimizing the impact of imperfections in synthetic data on the generated head images. Extensive experiments show that our model outperforms existing methods in terms of realism, expressiveness, and 3D consistency. Our code, synthetic datasets, and pre-trained models will be released in our project page: https://ustc3dv.github.io/Learn2Control/

cs.GR↗

arXiv2025

CHOP: Mobile Operating Assistant with Constrained High-frequency Optimized Subtask Planning

The advancement of visual language models (VLMs) has enhanced mobile device operations, allowing simulated human-like actions to address user requirements. Current VLM-based mobile operating assistants can be structured into three levels: task, subtask, and action. The subtask level, linking high-level goals with low-level executable actions, is crucial for task completion but faces two challenges: ineffective subtasks that lower-level agent cannot execute and inefficient subtasks that fail to contribute to the completion of the higher-level task. These challenges stem from VLM's lack of experience in decomposing subtasks within GUI scenarios in multi-agent architecture. To address these, we propose a new mobile assistant architecture with constrained high-frequency o}ptimized planning (CHOP). Our approach overcomes the VLM's deficiency in GUI scenarios planning by using human-planned subtasks as the basis vector. We evaluate our architecture in both English and Chinese contexts across 20 Apps, demonstrating significant improvements in both effectiveness and efficiency. Our dataset and code is available at https://github.com/Yuqi-Zhou/CHOP

cs.AI↗

arXiv2025

Shape from Semantics: 3D Shape Generation from Multi-View Semantics

Existing 3D reconstruction methods utilize guidances such as 2D images, 3D point clouds, shape contours and single semantics to recover the 3D surface, which limits the creative exploration of 3D modeling. In this paper, we propose a novel 3D modeling task called ``Shape from Semantics'', which aims to create 3D models whose geometry and appearance are consistent with the given text semantics when viewed from different views. The reconstructed 3D models incorporate more than one semantic elements and are easy for observers to distinguish. We adopt generative models as priors and disentangle the connection between geometry and appearance to solve this challenging problem. Specifically, we propose Local Geometry-Aware Distillation (LGAD), a strategy that employs multi-view normal-depth diffusion priors to complete partial geometries, ensuring realistic shape generation. We also integrate view-adaptive guidance scales to enable smooth semantic transitions across views. For appearance modeling, we adopt physically based rendering to generate high-quality material properties, which are subsequently baked into fabricable meshes. Extensive experimental results demonstrate that our method can generate meshes with well-structured, intricately detailed geometries, coherent textures, and smooth transitions, resulting in visually appealing 3D shape designs. Project page: https://shapefromsemantics.github.io

cs.CV↗