arXiv subjects

Bowen Liu

Publications and source records attributed to Bowen Liu.

At least 55 records · Page 3Linked to original sources

arXiv2026

EyeWorld: A Generative World Model of Ocular State and Dynamics

Ophthalmic decision-making depends on subtle lesion-scale cues interpreted across multimodal imaging and over time, yet most medical foundation models remain static and degrade under modality and acquisition shifts. Here we introduce EyeWorld, a generative world model that conceptualizes the eye as a partially observed dynamical system grounded in clinical imaging. EyeWorld learns an observation-stable latent ocular state shared across modalities, unifying fine-grained parsing, structure-preserving cross-modality translation and quality-robust enhancement within a single framework. Longitudinal supervision further enables time-conditioned state transitions, supporting forecasting of clinically meaningful progression while preserving stable anatomy. By moving from static representation learning to explicit dynamical modeling, EyeWorld provides a unified approach to robust multimodal interpretation and prognosis-oriented simulation in medicine.

cs.CV↗

arXiv2026

EyeAgent: An Agentic AI System for Multimodal Clinical Decision Support in Ophthalmology

Artificial intelligence has shown promise in medical imaging, yet most existing systems lack flexibility, interpretability, and adaptability - challenges especially pronounced in ophthalmology, where diverse imaging modalities are essential. We present EyeAgent, the first agentic AI framework for comprehensive and interpretable clinical decision support in ophthalmology. Using a large language model (DeepSeek-V3) as its central reasoning engine, EyeAgent interprets user queries and dynamically orchestrates 53 validated ophthalmic tools across 23 imaging modalities for diverse tasks including classification, segmentation, detection, image/report generation, and quantitative analysis. Stepwise ablation analysis demonstrated a progressive improvement in diagnostic accuracy, rising from a baseline of 69.71% (using only 5 general tools) to 80.79% when the full suite of 53 specialized tools was integrated. In an expert rating study on 200 real-world clinical cases, EyeAgent achieved 93.7% tool selection accuracy and received expert ratings of more than 88% across accuracy, completeness, safety, reasoning, and interpretability. In human-AI collaboration, EyeAgent matched or exceeded the performance of senior ophthalmologists and, when used as an assistant, improved overall diagnostic accuracy by 18.51% and report quality scores by 19%, with the greatest benefit observed among junior ophthalmologists. These findings establish EyeAgent as a scalable and trustworthy AI framework for ophthalmology and provide a blueprint for modular, multimodal, and clinically aligned next-generation AI systems.

cs.HC↗

arXiv2026

QCAgent: An agentic framework for quality-controllable pathology report generation from whole slide image

Recent methods for pathology report generation from whole-slide image (WSI) are capable of producing slide-level diagnostic descriptions but fail to ground fine-grained statements in localized visual evidence. Furthermore, they lack control over which diagnostic details to include and how to verify them. Inspired by emerging agentic analysis paradigms and the diagnostic workflow of pathologists,who selectively examine multiple fields of view, we propose QCAgent, an agentic framework for quality-controllable WSI report generation. The core innovations of this framework are as follows: (i) it incorporates a customized critique mechanism guided by a user-defined checklist specifying required diagnostic details and constraints; (ii) it re-identifies informative regions in the WSI based on the critique feedback and text-patch semantic retrieval, a process that iteratively enriches and reconciles the report. Experiments demonstrate that by making report requirements explicitly prompt-defined, constraint-aware, and verifiable through evidence-grounded refinement, QCAgent enables controllable generation of clinically meaningful and high-coverage pathology reports from WSI.

cs.CV↗

arXiv2026

NPG-Muse: Scaling Long Chain-of-Thought Reasoning with NP-Hard Graph Problems

Reasoning Large Language Models (RLLMs) have recently achieved remarkable progress on complex reasoning tasks, largely enabled by their long chain-of-thought (Long CoT) capabilities. However, developing these Long CoT behaviors relies heavily on post-training with high-quality datasets, which are typically costly and human-curated (e.g., mathematics and code), leaving scalable alternatives unexplored. In this work, we introduce NP-hard (NPH) graph problems as a novel synthetic training corpus, as they inherently require deep reasoning, extensive exploration, and reflective strategies, which are the core characteristics of Long CoT reasoning. Building on this insight, we develop a two-stage post-training framework: (i) Long-CoT Supervised Fine-Tuning (SFT) on rejection-sampled NPH graph instances, which substantially enhances reasoning depth, and (ii) Reinforcement Learning (RL) with a fine-grained reward design, which sharpens reasoning efficiency. The resulting NPG-Muse-series models exhibit substantially enhanced Long CoT reasoning capabilities, achieving consistent gains across mathematics, coding, logical, and graph reasoning benchmarks. NPG-Muse-7B even surpasses QwQ-32B on NPH graph problems in both accuracy and reasoning efficiency. These results position NPH graph problems as an effective and scalable resource for advancing Long CoT reasoning in LLM post-training. Our implementation is available at https://github.com/littlewyy/NPG-Muse.

cs.CL↗

arXiv2026

AlgBench: To What Extent Do Large Reasoning Models Understand Algorithms?

Reasoning ability has become a central focus in the advancement of Large Reasoning Models (LRMs). Although notable progress has been achieved on several reasoning benchmarks such as MATH500 and LiveCodeBench, existing benchmarks for algorithmic reasoning remain limited, failing to answer a critical question: Do LRMs truly master algorithmic reasoning? To answer this question, we propose AlgBench, an expert-curated benchmark that evaluates LRMs under an algorithm-centric paradigm. AlgBench consists of over 3,000 original problems spanning 27 algorithms, constructed by ACM algorithmic experts and organized under a comprehensive taxonomy, including Euclidean-structured, non-Euclidean-structured, non-optimized, local-optimized, global-optimized, and heuristic-optimized categories. Empirical evaluations on leading LRMs (e.g., Gemini-3-Pro, DeepSeek-v3.2-Speciale and GPT-o3) reveal substantial performance heterogeneity: while models perform well on non-optimized tasks (up to 92%), accuracy drops sharply to around 49% on globally optimized algorithms such as dynamic programming. Further analysis uncovers \textbf{strategic over-shifts}, wherein models prematurely abandon correct algorithmic designs due to necessary low-entropy tokens. These findings expose fundamental limitations of problem-centric reinforcement learning and highlight the necessity of an algorithm-centric training paradigm for robust algorithmic reasoning.

cs.AI↗

arXiv2025

New Theoretical Insights and Algorithmic Solutions for Reconstructing Score Sequences from Tournament Score Sets

The score set of a tournament is defined as the set of its distinct out-degrees. In 1978, Reid proposed the conjecture that for any set of nonnegative integers $D$, there exists a tournament $T$ with a degree set $D$. In 1989, Yao presented an arithmetical proof of the conjecture, but a general polynomial-time construction algorithm is not known. This paper proposes a necessary and sufficient condition and a separate necessary condition, based on the existing Landau's theorem for the problem of reconstructing score sequences from score sets of tournament graphs. The necessary condition introduces a structured set that enables the use of group-theoretic techniques, offering not only a framework for solving the reconstruction problem but also a new perspective for approaching similar problems. In particular, the same theoretical approach can be extended to reconstruct valid score sets given constraints on the frequency of distinct scores in tournaments. Based on these conditions, we have developed three algorithms that demonstrate the practical utility of our framework: a polynomial-time algorithm and a scalable algorithm for reconstructing score sequences, and a polynomial-time network-building method that finds all possible score sequences for a given score set. Moreover, the polynomial-time algorithm for reconstructing the score sequence of a tournament for a given score set can be used to verify Reid's conjecture. These algorithms have practical applications in sports analysis, ranking prediction, and machine learning tasks such as learning-to-rank models and data imputation, where the reconstruction of partial rankings or sequences is essential for recommendation systems and anomaly detection.

cs.DS↗

arXiv2025

Understanding What Is Not Said:Referring Remote Sensing Image Segmentation with Scarce Expressions

Referring Remote Sensing Image Segmentation (RRSIS) aims to segment instances in remote sensing images according to referring expressions. Unlike Referring Image Segmentation on general images, acquiring high-quality referring expressions in the remote sensing domain is particularly challenging due to the prevalence of small, densely distributed objects and complex backgrounds. This paper introduces a new learning paradigm, Weakly Referring Expression Learning (WREL) for RRSIS, which leverages abundant class names as weakly referring expressions together with a small set of accurate ones to enable efficient training under limited annotation conditions. Furthermore, we provide a theoretical analysis showing that mixed-referring training yields a provable upper bound on the performance gap relative to training with fully annotated referring expressions, thereby establishing the validity of this new setting. We also propose LRB-WREL, which integrates a Learnable Reference Bank (LRB) to refine weakly referring expressions through sample-specific prompt embeddings that enrich coarse class-name inputs. Combined with a teacher-student optimization framework using dynamically scheduled EMA updates, LRB-WREL stabilizes training and enhances cross-modal generalization under noisy weakly referring supervision. Extensive experiments on our newly constructed benchmark with varying weakly referring data ratios validate both the theoretical insights and the practical effectiveness of WREL and LRB-WREL, demonstrating that they can approach or even surpass models trained with fully annotated referring expressions.

eess.IV↗

arXiv2025

Mask What Matters: Controllable Text-Guided Masking for Self-Supervised Medical Image Analysis

The scarcity of annotated data in specialized domains such as medical imaging presents significant challenges to training robust vision models. While self-supervised masked image modeling (MIM) offers a promising solution, existing approaches largely rely on random high-ratio masking, leading to inefficiency and poor semantic alignment. Moreover, region-aware variants typically depend on reconstruction heuristics or supervised signals, limiting their adaptability across tasks and modalities. We propose Mask What Matters, a controllable text-guided masking framework for self-supervised medical image analysis. By leveraging vision-language models for prompt-based region localization, our method flexibly applies differentiated masking to emphasize diagnostically relevant regions while reducing redundancy in background areas. This controllable design enables better semantic alignment, improved representation learning, and stronger cross-task generalizability. Comprehensive evaluation across multiple medical imaging modalities, including brain MRI, chest CT, and lung X-ray, shows that Mask What Matters consistently outperforms existing MIM methods (e.g., SparK), achieving gains of up to +3.1 percentage points in classification accuracy, +1.3 in box average precision (BoxAP), and +1.1 in mask average precision (MaskAP) for detection. Notably, it achieves these improvements with substantially lower overall masking ratios (e.g., 40\% vs. 70\%). This work demonstrates that controllable, text-driven masking can enable semantically aligned self-supervised learning, advancing the development of robust vision models for medical image analysis.

cs.CV↗

arXiv2025

Phase Noise Tolerance for Low-Pilot-Overhead OFDM Terahertz Links Beyond 64-QAM

THz wireless communications have garnered significant attention due to their unprecedented data rates enabled by the abundant untapped spectrum. However, advanced modulation formats beyond 64-QAM remain largely unexplored, as phase errors introduced during up/down-conversion severely limit system performance. Particularly, OFDM transmission is highly susceptible to aggravated ICI induced by phase noise, undermining the orthogonality of subcarriers. While PLLs and pilot-assisted compensation can mitigate phase errors, excessive pilot overhead compromises spectral efficiency and energy consumption, and white phase noise remains unrecoverable. Therefore, quantifying phase noise tolerance is essential for practical physical layer protocols. Here, we reveal the impact of phase noise in a 64-QAM, 2048-subcarrier OFDM THz transmission system. 3σ-error estimation is proposed to quantify phase noise tolerance, indicating an intuitive EVM threshold of approximately 5%. This threshold further delineates the trade-offs among phase noise levels, SNR requirements, and pilot overhead. Moreover, by benchmarking representative oscillators with distinct phase noise spectra, microring resonators (MRRs) are identified as indispensable enablers for low-pilot-overhead OFDM THz links operating beyond 64-QAM.

physics.optics↗

arXiv2025

Quantifying Phase Noise Tolerance for Single-Carrier M-QAM Terahertz Wireless Communications with Advantages of Photonic Approaches

Terahertz wireless communications offer abundant untapped spectrum and are regarded as a promising playground for next-generation high-throughput links. Yet oscillator phase noise becomes the dominant impairment at such high frequencies, severely limiting the reliability of high-order QAM transmission. While photonic approaches, such as microcombs, are known to realize ultralow phase noise, the quantitative level of suppression required to sustain reliable high-order QAM transmission has not been clarified. Here, phase noise is reconstructed from measured spectra and embedded into a single-carrier link model to evaluate its impact. Distinct distortion mechanisms are identified, with slow common phase error and instantaneous phase jitter, where the latter remains as the residual impairment after carrier phase recovery. We further adopt the 3σ error criterion, which maps residual distortions onto the constellation, providing a clear and practical indicator of system robustness. The results indicate that modest improvements in oscillator stability translate into significant BER gains without proportional power increase. These findings provide intuitive tolerance of phase noise in M-QAM systems and emphasize the importance of integrating low-noise photonic oscillators such as microcombs.

physics.optics↗

arXiv2025

Photonics-Aware Planning-Guided Automated Electrical Routing for Large-Scale Active Photonic Integrated Circuits

The rising demand for AI training and inference, as well as scientific computing, combined with stringent latency and energy budgets, is driving the adoption of integrated photonics for computing, sensing, and communications. As active photonic integrated circuits (PICs) scale in device count and functional heterogeneity, physical implementation by manual scripting and ad-hoc edits is no longer tenable. This creates an immediate need for an electronic-photonic design automation (EPDA) stack in which physical design automation is a core capability. However, there is currently no end-to-end fully automated routing flow that coordinates photonic waveguides and on-chip metal interconnect. Critically, available digital VLSI and analog/custom routers are not directly applicable to PIC metal routing due to a lack of customization to handle constraints induced by photonic devices and waveguides. We present, to our knowledge, the first end-to-end routing framework for large-scale active PICs that jointly addresses waveguides and metal wires within a unified flow. We introduce a physically-aware global planner that generates congestion- and crossing-aware routing guides while explicitly accounting for the placement of photonic components and waveguides. We further propose a sequence-consistent track assignment and a soft guidance-assisted detailed routing to speed up the routing process with significantly optimized routability and via usage. Evaluated on various large PIC designs, our router delivers fast, high-quality active PIC routing solutions with fewer vias, lower congestion, and competitive runtime relative to manual and existing VLSI router baselines; on average it reduce via count by ~99%, user-specified design rule violation by ~98%, and runtime by 17x, establishing a practical foundation for EPDA at system scale.

physics.optics↗

arXiv2025

Short-wave infrared broadband up-conversion imaging by using a noncritical phase matched bulk KTiOPO$_4$ crystal

Compared to cryogenically cooled conventional detectors, up-conversion detection enables efficient room-temperature short-wave infrared (SWIR) imaging. Although quasi-phase-matching (QPM) in periodically poled crystals offers advantages, the small crystal aperture (typically 1 mm$\times$3 mm) limits resolution. Non-poled crystals enable larger apertures but suffer walk-off aberrations. This work overcomes these limitations by using a noncritical phase matched (NCPM) KTiOPO$_4$ crystal (6 mm$\times$7 mm aperture, 0.5 mm length). Results show resolutions 6$\times$ and 2$\times$ higher than periodically poled crystals in orthogonal directions, with broad conversion band (1.3-2.2 $μ$m) covering biological and atmospheric windows. The absence of walk-off ensures better image fidelity in up-conversion process. This study presents the first comprehensive characterization of NCPM-based broadband up-conversion imaging, demonstrating performance at the theoretical resolution limit while circumventing drawbacks inherent in alternative up-conversion schemes and conventional detectors.

physics.optics↗

arXiv2025

LESER: Learning to Expand via Search Engine-feedback Reinforcement in e-Commerce

User queries in e-commerce search are often vague, short, and underspecified, making it difficult for retrieval systems to match them accurately against structured product catalogs. This challenge is amplified by the one-to-many nature of user intent, where a single query can imply diverse and competing needs. Existing methods, including neural query expansion and prompting-based LLM approaches, fall short in real-world settings: they struggle to capture nuanced user intent, often generate outputs that violate platform constraints, and rely on workflows that are difficult to scale in production. We propose Learning to Expand via Search Engine-feedback Reinforcement (LESER), a novel framework that fine-tunes a context-aware LLM using real-time search engine feedback as supervision. LESER formulates query expansion as a retrieval optimization task and leverages Group Relative Policy Optimization to learn directly from relevance and coverage metrics. LESER is trained to reason over search results and produce high quality query expansions that align with platform rules and retrieval objectives. We evaluate LESER on large-scale, real-world e-commerce datasets, demonstrating substantial improvements in both offline and online settings. Our results show that LESER not only enhances semantic coverage and retrieval relevance but also delivers measurable gains in user engagement, making it a practical and scalable solution for modern search systems.

cs.IR↗

arXiv2025

Repetition-Rate-Difference Tunable Dual-Comb Fiber Laser Using Bidirectional Lyot filtering

Single cavity dual-comb fiber lasers adopting different multiplexing configurations are benefited from the natures of common-mode noise suppression and superior coherence. Particularly, repetition-rate tunable dual-combs enable non-ambiguous ranging and aliasing-free spectroscopy. However, their sampling rate and spectral resolution is severely restricted by the mechanical delay lines. In a previous work, as rapid as 500 kHz/s tuning rate was realized to address this issue, while the minimum comb frequency difference remained large under the inaccuracy of mechanical DLL. In this work, a dual-comb prototype incorporated with a thermally controlled bidirectional lyot filter is demonstrated with 870-times enhanced tuning precision compared with mechanical schemes. Linear correlation between temperature and repetition-rate-difference of this tuning mechanism is revealed. We achieve a tuning efficiency of 4.4 Hz/°C and a control accuracy of 0.44 Hz/K, denoting a significant advance in operating Hz-scale differential comb lines. This design offers an optimal playground for extending non-ambiguous distance in dead-zone-free dual-comb ranging and eliminating aliasing in spectroscopy.

physics.optics↗

arXiv2025

Impacts of Phase Noise on M-ary QAM THz Wireless Communications

THz technology is positioned as a key enabler for next-generation wireless links due to the extensive untapped bandwidth and inherent compatibility with silicon photonics. Here, the phase noise of THz sources is modeled to characterized its impacts on a QAM-based wireless communication system. Particularly, common and instantaneous phase errors are unveiled and presented via constellation diagrams. In addition, tolerance of phase noise under different M-ary modulation formats is explored. Meanwhile, a linear correlation between instantaneous component and EVM is revealed to define the error-free boundary, where microcomb-driven scheme is highlighted as a highly promising candidate for error-resilient THz links under advanced M-QAM formats. The results intuitively illustrate the impacts of phase noise and offer practical insights for refining technical details on physical-layer protocols.

physics.optics↗

arXiv2025

Identifying Ring Galaxies in DESI Legacy Imaging Surveys Using Machine Learning Methods

The formation and evolution of ring structures in galaxies are crucial for understanding the nature and distribution of dark matter, galactic interactions, and the internal secular evolution of galaxies. However, the limited number of existing ring galaxy catalogs has constrained deeper exploration in this field. To address this gap, we introduce a two-stage binary classification model based on the Swin Transformer architecture to identify ring galaxies from the DESI Legacy Imaging Surveys. This model first selects potential candidates and then refines them in a second stage to improve classification accuracy. During model training, we investigated the impact of imbalanced datasets on the performance of the two-stage model. We experimented with various model combinations applied to the datasets of the DESI Legacy Imaging Surveys DR9, processing a total of 573,668 images with redshifts ranging from z_spec = 0.01-0.20 and magr <17.5. After applying the two-stage filtering and conducting visual inspections, the overall Precision of the models exceeded 64.87%, successfully identifying a total of 8052 newly discovered ring galaxies. With our catalog, the forthcoming spectroscopic data from DESI will facilitate a more comprehensive investigation into the formation and evolution of ring galaxies.

astro-ph.GA↗

arXiv2025

APTOS-2024 challenge report: Generation of synthetic 3D OCT images from fundus photographs

Optical Coherence Tomography (OCT) provides high-resolution, 3D, and non-invasive visualization of retinal layers in vivo, serving as a critical tool for lesion localization and disease diagnosis. However, its widespread adoption is limited by equipment costs and the need for specialized operators. In comparison, 2D color fundus photography offers faster acquisition and greater accessibility with less dependence on expensive devices. Although generative artificial intelligence has demonstrated promising results in medical image synthesis, translating 2D fundus images into 3D OCT images presents unique challenges due to inherent differences in data dimensionality and biological information between modalities. To advance generative models in the fundus-to-3D-OCT setting, the Asia Pacific Tele-Ophthalmology Society (APTOS-2024) organized a challenge titled Artificial Intelligence-based OCT Generation from Fundus Images. This paper details the challenge framework (referred to as APTOS-2024 Challenge), including: the benchmark dataset, evaluation methodology featuring two fidelity metrics-image-based distance (pixel-level OCT B-scan similarity) and video-based distance (semantic-level volumetric consistency), and analysis of top-performing solutions. The challenge attracted 342 participating teams, with 42 preliminary submissions and 9 finalists. Leading methodologies incorporated innovations in hybrid data preprocessing or augmentation (cross-modality collaborative paradigms), pre-training on external ophthalmic imaging datasets, integration of vision foundation models, and model architecture improvement. The APTOS-2024 Challenge is the first benchmark demonstrating the feasibility of fundus-to-3D-OCT synthesis as a potential solution for improving ophthalmic care accessibility in under-resourced healthcare settings, while helping to expedite medical research and clinical applications.

cs.CV↗

arXiv2025

"Salt is the Soul of Hakka Baked Chicken": Reimagining Traditional Chinese Culinary ICH for Modern Contexts Without Losing Tradition

Intangible Cultural Heritage (ICH) like traditional culinary practices face increasing pressure to adapt to globalization while maintaining their cultural authenticity. Centuries-old traditions in Chinese cuisine are subject to rapid changes for adaptation to contemporary tastes and dietary preferences. The preservation of these cultural practices requires approaches that can enable ICH practitioners to reimagine and recreate ICH for modern contexts. To address this, we created workshops where experienced practitioners of traditional Chinese cuisine co-created recipes using GenAI tools and realized the dishes. We found that GenAI inspired ICH practitioners to innovate recipes based on traditional workflows for broader audiences and adapt to modern dining contexts. However, GenAI-inspired co-creation posed challenges in maintaining the accuracy of original ICH workflows and preserving traditional flavors in the culinary outcomes. This study offers implications for designing human-AI collaborative processes for safeguarding and enhancing culinary ICH.

cs.HC↗