arXiv subjects

Ruijie Yang

Publications and source records attributed to Ruijie Yang.

At least 19 recordsLinked to original sources

arXiv2026

Filtrations on D-modules and multiplicities of roots of Bernstein-Sato polynomials

In this paper, we relate multiplicities of Bernstein--Sato-type polynomials with respect to a holomorphic function f to several singularity invariants. First, we introduce certain ``mod'' b-functions and show that they characterize the weight filtration on the localization of a simple regular holonomic D-module along f, and we provide an algorithm to compute them. Second, we show that they can be approximated by the multiplicities of roots of power b-functions (the b-functions with respect to powers of f). Further, we give a sharp upper bound for the Hodge level of elements given by a certain sum of such multiplicities. Next, we give an effective asymptotic solution to the Gelfand problem by determining an explicit threshold after which every integer shift of a root of b_f(s) is a pole of the Archimedean zeta function of f. We also show that the order of these poles is equal to the nilpotency index of the logarithmic monodromy operator, which we further express as the limit of the multiplicities of roots of power b-functions. We define several filtrations, relating them to the weight and Hodge filtrations, based upon which we leave some open questions that we address in the affirmative in the case when f has a homogeneous isolated singularity, or it is a hyperplane arrangement, or it is a semi-invariant on a spherical variety. We give several immediate applications to our results, including a positive answer to a question of Torelli assuming the hypersurface has log canonical singularities: 1/f lies in the intersection complex of the hypersurface of f if and only if -1 is a simple root of b_f(s).

math.AG↗

arXiv2026

From Generalist to Specialist: A Context-Fusion Framework for Endoscopic Polyp Reporting with a Frozen VLM

Reliable endoscopic polyp reporting requires integrating quantitative lesion sizing, standardized Paris classification, and clinically meaningful morphological description within a single record. General-purpose vision-language models (VLMs) offer a unified interface for image understanding and report generation. Existing specialization strategies, however, typically rely on task-specific models or model-weight adaptation, leaving unresolved how to introduce reliable specialist knowledge while preserving both this unified interface and the VLM's pretrained capabilities. We introduce a context-fusion framework that specializes a frozen general-purpose VLM through both implicit instruction context and explicit transduction context without modifying its pretrained weights. Specifically, a self-supervised polyp encoder retrieves related image-report pairs as explicit, query-specific evidence, while learned continuous specialist tokens provide implicit instruction context shared across cases. Experiments were conducted on 2,056 expert-annotated public endoscopic images. We compared the framework with general-purpose VLMs, task-specific predictors, and weight-adaptation methods to assess specialist performance, unified reporting, and adaptation efficiency. Across numerical, categorical, and report-generation metrics, the proposed framework substantially improved direct frozen-VLM inference and achieved the strongest overall performance among the evaluated methods. It added trainable parameters equal to only 0.006% of the frozen VLM's parameter count. When the top-1 retrieved case carried the correct target category, our framework corrected 70.5% of the errors made by a weight-adaptation baseline. These findings support the context-fusion framework as a lightweight and effective strategy for specialist adaptation of a frozen VLM.

cs.AI↗

arXiv2026

A report-grounded vision-language foundation model for colonoscopy from 280000 routine reports

Vision-language models remain underused in colonoscopy despite the rich expert descriptions recorded in routine reports. These reports document lesion appearance, size and location but summarise entire procedures rather than caption individual frames, leaving clinical findings only weakly linked to the corresponding images. Here we develop EndoCLIP, a colonoscopy vision-language foundation model trained on 125,756 lesion-level image-text pairs progressively recovered from 280,476 routine colonoscopy records. Across lesion-level image-text retrieval, structured report generation and six multi-centre clinical classification tasks, EndoCLIP outperforms general-purpose and biomedical vision-language encoders in both zero-shot and linear-probe settings. On benign-versus-malignant classification, its linear probe approaches the performance of expert readers in a blinded study involving 12 endoscopists. These results suggest that recovering finding-to-frame correspondence can transform routine documentation into scalable supervision, enabling clinical targets to be specified in language rather than separately annotated for each task.

cs.AI↗

arXiv2026

Multivariate V-filtrations and the Strong Monodromy Conjecture for hyperplane arrangements

In this work, we develop a new theory of multivariate V-filtration on D-modules along a simple normal crossing divisor and relate it with Sabbah's multi-filtration. We establish several new structural results and relate them with the Hodge filtration on free-monodromic local systems from geometric representation theory. As an illustrative application, we give a conceptual and very quick proof of the Strong Monodromy Conjecture and its multivariate generalisation for hyperplane arrangements. Along the way, we confirm both the n/d-conjecture of Budur--Musta\c{t}\u{a}--Teitler and its multivariate form due to Budur.

math.AG↗

arXiv2026

Development and multi-center evaluation of domain-adapted speech recognition for human-AI teaming in real-world gastrointestinal endoscopy

Automatic speech recognition (ASR) is a critical interface for human-AI interaction in gastrointestinal endoscopy, yet its reliability in real-world clinical settings is limited by domain-specific terminology and complex acoustic conditions. Here, we present EndoASR, a domain-adapted ASR system designed for real-time deployment in endoscopic workflows. We develop a two-stage adaptation strategy based on synthetic endoscopy reports, targeting domain-specific language modeling and noise robustness. In retrospective evaluation across six endoscopists, EndoASR substantially improves both transcription accuracy and clinical usability, reducing character error rate (CER) from 20.52% to 14.14% and increasing medical term accuracy (Med ACC) from 54.30% to 87.59%. In a prospective multi-center study spanning five independent endoscopy centers, EndoASR demonstrates consistent generalization under heterogeneous real-world conditions. Compared with the baseline Paraformer model, CER is reduced from 16.20% to 14.97%, while Med ACC is improved from 61.63% to 84.16%, confirming its robustness in practical deployment scenarios. Notably, EndoASR achieves a real-time factor (RTF) of 0.005, significantly faster than Whisper-large-v3 (RTF 0.055), while maintaining a compact model size of 220M parameters, enabling efficient edge deployment. Furthermore, integration with large language models demonstrates that improved ASR quality directly enhances downstream structured information extraction and clinician-AI interaction. These results demonstrate that domain-adapted ASR can serve as a reliable interface for human-AI teaming in gastrointestinal endoscopy, with consistent performance validated across multi-center real-world clinical settings.

cs.CL↗

arXiv2025

EndoFinder: Online Lesion Retrieval for Explainable Colorectal Polyp Diagnosis Leveraging Latent Scene Representations

Colorectal cancer (CRC) remains a leading cause of cancer-related mortality, underscoring the importance of timely polyp detection and diagnosis. While deep learning models have improved optical-assisted diagnostics, they often demand extensive labeled datasets and yield "black-box" outputs with limited interpretability. In this paper, we propose EndoFinder, an online polyp retrieval framework that leverages multi-view scene representations for explainable and scalable CRC diagnosis. First, we develop a Polyp-aware Image Encoder by combining contrastive learning and a reconstruction task, guided by polyp segmentation masks. This self-supervised approach captures robust features without relying on large-scale annotated data. Next, we treat each polyp as a three-dimensional "scene" and introduce a Scene Representation Transformer, which fuses multiple views of the polyp into a single latent representation. By discretizing this representation through a hashing layer, EndoFinder enables real-time retrieval from a compiled database of historical polyp cases, where diagnostic information serves as interpretable references for new queries. We evaluate EndoFinder on both public and newly collected polyp datasets for re-identification and pathology classification. Results show that EndoFinder outperforms existing methods in accuracy while providing transparent, retrieval-based insights for clinical decision-making. By contributing a novel dataset and a scalable, explainable framework, our work addresses key challenges in polyp diagnosis and offers a promising direction for more efficient AI-driven colonoscopy workflows. The source code is available at https://github.com/ku262/EndoFinder-Scene.

cs.IR↗

arXiv2025

Endo-CLIP: Progressive Self-Supervised Pre-training on Raw Colonoscopy Records

Pre-training on image-text colonoscopy records offers substantial potential for improving endoscopic image analysis, but faces challenges including non-informative background images, complex medical terminology, and ambiguous multi-lesion descriptions. We introduce Endo-CLIP, a novel self-supervised framework that enhances Contrastive Language-Image Pre-training (CLIP) for this domain. Endo-CLIP's three-stage framework--cleansing, attunement, and unification--addresses these challenges by (1) removing background frames, (2) leveraging large language models to extract clinical attributes for fine-grained contrastive learning, and (3) employing patient-level cross-attention to resolve multi-polyp ambiguities. Extensive experiments demonstrate that Endo-CLIP significantly outperforms state-of-the-art pre-training methods in zero-shot and few-shot polyp detection and classification, paving the way for more accurate and clinically relevant endoscopic analysis.

cs.CV↗

arXiv2025

Filtrations of D-modules along semi-invariant functions

Given a smooth algebraic variety X with an action of a connected reductive linear algebraic group G, and an equivariant D-module M, we study the G-decompositions of the associated V-, Hodge, and weight filtrations. If M is the localization of a D-module S underlying a pure twistor D-module (e.g. when S is simple) along a semi-invariant function, we determine the weight level of any element in an irreducible isotypic component of M in terms of multiplicities of roots of b-functions. If S underlies a pure Hodge module, we show that the Hodge level is governed by the degrees of another class of polynomials, also expressible in terms of b-functions. As an application, if X is an affine spherical variety, we describe these filtrations representation-theoretically in terms of roots of b-functions, and compute all higher multiplier and Hodge ideals associated with semi-invariant functions. Examples include the spaces of general, skew-symmetric, and symmetric matrices, as well as the Freudenthal cubic on the fundamental representation of E_6.

math.AG↗

arXiv2025

On the Hodge and V-filtrations of mixed Hodge modules

In this paper, we prove a Beilinson-type formula for the V-filtration of Kashiwara and Malgrange on a complex mixed Hodge module, using Hodge filtrations on the localization. Our formula expresses the V-filtration as the filtered D-module underlying a pro-mixed Hodge module. We apply this to the theory of higher multiplier and Hodge ideals. Our first result shows that higher multiplier ideals can be obtained directly from Hodge ideals by taking a suitable limit. As a corollary, we deduce that Hodge ideals are left semi-continuous if and only if they coincide with higher multiplier ideals, thereby improving results of Saito and Musta\c{t}\u{a}--Popa and resolving a folklore question. We further prove a birational transformation formula for higher multiplier ideals, generalising the classical formula for multiplier ideals and answering a question of Schnell and the second author. Finally, we provide very quick proofs of the main vanishing theorems for Hodge ideals, and strengthen a result of B. Chen.

math.AG↗

arXiv2024

Generating Editable Head Avatars with 3D Gaussian GANs

Generating animatable and editable 3D head avatars is essential for various applications in computer vision and graphics. Traditional 3D-aware generative adversarial networks (GANs), often using implicit fields like Neural Radiance Fields (NeRF), achieve photorealistic and view-consistent 3D head synthesis. However, these methods face limitations in deformation flexibility and editability, hindering the creation of lifelike and easily modifiable 3D heads. We propose a novel approach that enhances the editability and animation control of 3D head avatars by incorporating 3D Gaussian Splatting (3DGS) as an explicit 3D representation. This method enables easier illumination control and improved editability. Central to our approach is the Editable Gaussian Head (EG-Head) model, which combines a 3D Morphable Model (3DMM) with texture maps, allowing precise expression control and flexible texture editing for accurate animation while preserving identity. To capture complex non-facial geometries like hair, we use an auxiliary set of 3DGS and tri-plane features. Extensive experiments demonstrate that our approach delivers high-quality 3D-aware synthesis with state-of-the-art controllability. Our code and models are available at https://github.com/liguohao96/EGG3D.

cs.CV↗

arXiv2024

Archimedean zeta functions, singularities, and Hodge theory

We use Hodge theory to relate poles of the Archimedean zeta function $Z_f$ of a holomorphic function $f$ with several invariants of singularities. First, we prove that the largest nontrivial pole of $Z_f$ is the negative of the minimal exponent of $f$, whose order is determined by the multiplicity of the corresponding root of the Bernstein--Sato polynomial $b_f(s)$, resolving in a strong sense a question of Musta\c{t}\u{a}--Popa. This simultaneously generalizes a result of Loeser for isolated singularities and of Koll\'ar--Litchin for the log canonical threshold, and improves them by accounting for the multiplicity. On the other hand, we give an example of $f$ where a root of $b_f(s)$ is not a pole of $Z_f$, answering a question of Loeser from 1985 in the negative. As a byproduct, we give a positive answer to a question of Budur--Walther in the case of the minimal exponent. In general, we determine poles of $Z_f$ from the Hodge filtration on vanishing cycles, sharpening a result of Barlet. Finally, we obtain analytic descriptions of the $V$-filtration of Kashiwara and Malgrange, Hodge and higher multiplier ideals, addressing another question of Musta\c{t}\u{a}--Popa. The proofs mainly rely on a positivity property of the polarization on the lowest piece of the Hodge filtration on a complex Hodge module in the sense of Sabbah--Schnell.

math.AG↗

arXiv2024

Multi-modal Relation Distillation for Unified 3D Representation Learning

Recent advancements in multi-modal pre-training for 3D point clouds have demonstrated promising results by aligning heterogeneous features across 3D shapes and their corresponding 2D images and language descriptions. However, current straightforward solutions often overlook intricate structural relations among samples, potentially limiting the full capabilities of multi-modal learning. To address this issue, we introduce Multi-modal Relation Distillation (MRD), a tri-modal pre-training framework, which is designed to effectively distill reputable large Vision-Language Models (VLM) into 3D backbones. MRD aims to capture both intra-relations within each modality as well as cross-relations between different modalities and produce more discriminative 3D shape representations. Notably, MRD achieves significant improvements in downstream zero-shot classification tasks and cross-modality retrieval tasks, delivering new state-of-the-art performance.

cs.CV↗

arXiv2024

EndoFinder: Online Image Retrieval for Explainable Colorectal Polyp Diagnosis

Determining the necessity of resecting malignant polyps during colonoscopy screen is crucial for patient outcomes, yet challenging due to the time-consuming and costly nature of histopathology examination. While deep learning-based classification models have shown promise in achieving optical biopsy with endoscopic images, they often suffer from a lack of explainability. To overcome this limitation, we introduce EndoFinder, a content-based image retrieval framework to find the 'digital twin' polyp in the reference database given a newly detected polyp. The clinical semantics of the new polyp can be inferred referring to the matched ones. EndoFinder pioneers a polyp-aware image encoder that is pre-trained on a large polyp dataset in a self-supervised way, merging masked image modeling with contrastive learning. This results in a generic embedding space ready for different downstream clinical tasks based on image retrieval. We validate the framework on polyp re-identification and optical biopsy tasks, with extensive experiments demonstrating that EndoFinder not only achieves explainable diagnostics but also matches the performance of supervised classification models. EndoFinder's reliance on image retrieval has the potential to support diverse downstream decision-making tasks during real-time colonoscopy procedures.

cs.CV↗

arXiv2024

Leveraging Predicate and Triplet Learning for Scene Graph Generation

Scene Graph Generation (SGG) aims to identify entities and predict the relationship triplets \textit{\textless subject, predicate, object\textgreater } in visual scenes. Given the prevalence of large visual variations of subject-object pairs even in the same predicate, it can be quite challenging to model and refine predicate representations directly across such pairs, which is however a common strategy adopted by most existing SGG methods. We observe that visual variations within the identical triplet are relatively small and certain relation cues are shared in the same type of triplet, which can potentially facilitate the relation learning in SGG. Moreover, for the long-tail problem widely studied in SGG task, it is also crucial to deal with the limited types and quantity of triplets in tail predicates. Accordingly, in this paper, we propose a Dual-granularity Relation Modeling (DRM) network to leverage fine-grained triplet cues besides the coarse-grained predicate ones. DRM utilizes contexts and semantics of predicate and triplet with Dual-granularity Constraints, generating compact and balanced representations from two perspectives to facilitate relation recognition. Furthermore, a Dual-granularity Knowledge Transfer (DKT) strategy is introduced to transfer variation from head predicates/triplets to tail ones, aiming to enrich the pattern diversity of tail classes to alleviate the long-tail problem. Extensive experiments demonstrate the effectiveness of our method, which establishes new state-of-the-art performance on Visual Genome, Open Image, and GQA datasets. Our code is available at \url{https://github.com/jkli1998/DRM}

cs.CV↗

arXiv2024

AED-PADA:Improving Generalizability of Adversarial Example Detection via Principal Adversarial Domain Adaptation

Adversarial example detection, which can be conveniently applied in many scenarios, is important in the area of adversarial defense. Unfortunately, existing detection methods suffer from poor generalization performance, because their training process usually relies on the examples generated from a single known adversarial attack and there exists a large discrepancy between the training and unseen testing adversarial examples. To address this issue, we propose a novel method, named Adversarial Example Detection via Principal Adversarial Domain Adaptation (AED-PADA). Specifically, our approach identifies the Principal Adversarial Domains (PADs), i.e., a combination of features of the adversarial examples generated by different attacks, which possesses a large portion of the entire adversarial feature space. Subsequently, we pioneer to exploit Multi-source Unsupervised Domain Adaptation in adversarial example detection, with PADs as the source domains. Experimental results demonstrate the superior generalization ability of our proposed AED-PADA. Note that this superiority is particularly achieved in challenging scenarios characterized by employing the minimal magnitude constraint for the perturbations.

cs.CV↗

arXiv2023

Higher multiplier ideals

We associate a family of ideal sheaves to any Q-effective divisor on a complex manifold, called higher multiplier ideals, using the theory of mixed Hodge modules and V-filtrations. This family is indexed by two parameters, an integer indicating the Hodge level and a rational number, and these ideals admit a weight filtration. When the Hodge level is zero, they recover the usual multiplier ideals. We study the local and global properties of higher multiplier ideals systematically. In particular, we prove vanishing theorems and restriction theorems, provide criteria for the nontriviality, and introduce the center of minimal exponent (generalizing the notion of minimal log canonical center). The main idea is to exploit the global structure of the V-filtration along an effective divisor using the notion of twisted Hodge modules. As applications, we prove new cases of conjectures by Debarre, Casalaina-Martin and Grushevsky on singularities of theta divisors on principally polarized abelian varieties.

math.AG↗

arXiv2023

Common Knowledge Learning for Generating Transferable Adversarial Examples

This paper focuses on an important type of black-box attacks, i.e., transfer-based adversarial attacks, where the adversary generates adversarial examples by a substitute (source) model and utilize them to attack an unseen target model, without knowing its information. Existing methods tend to give unsatisfactory adversarial transferability when the source and target models are from different types of DNN architectures (e.g. ResNet-18 and Swin Transformer). In this paper, we observe that the above phenomenon is induced by the output inconsistency problem. To alleviate this problem while effectively utilizing the existing DNN models, we propose a common knowledge learning (CKL) framework to learn better network weights to generate adversarial examples with better transferability, under fixed network architectures. Specifically, to reduce the model-specific features and obtain better output distributions, we construct a multi-teacher framework, where the knowledge is distilled from different teacher architectures into one student network. By considering that the gradient of input is usually utilized to generated adversarial examples, we impose constraints on the gradients between the student and teacher models, to further alleviate the output inconsistency problem and enhance the adversarial transferability. Extensive experiments demonstrate that our proposed work can significantly improve the adversarial transferability.

cs.LG↗

arXiv2023

Deep Reinforcement Learning for Beam Angle Optimization of Intensity-Modulated Radiation Therapy

Objective: Intensity-modulated radiation therapy (IMRT) beam angle optimization (BAO) is a challenging combinatorial optimization problem that is NP-hard. In this study, we aim to develop a personalized BAO algorithm for IMRT that improves the quality of the final treatment. Methods: To improve the quality of IMRT treatment planning, we propose a deep reinforcement learning (DRL)-based approach for IMRT BAO. We consider the task as a sequential decision-making problem and formulate it as a Markov Decision Process. To facilitate the training process, a 3D-Unet is designed to predict the dose distribution for the different number of beam angles, ranging from 1 to 9, to simulate the IMRT environment. By leveraging the simulation model, double deep-Q network (DDQN) and proximal policy optimization (PPO) are used to train agents to select the personalized beam angle sequentially within a few seconds. Results: The treatment plans with beam angles selected by DRL outperform those with clinically used evenly distributed beam angles. For DDQN, the overall average improvement of the CIs is 0.027, 0.032, and 0.03 for 5, 7, and 9 beam angles respectively. For PPO, the overall average improvement of CIs is 0.045, 0.051, and 0.025 for 5, 7, and 9 beam angles respectively. Conclusion: The proposed DRL-based beam angle selection strategy can generate personalized beam angles within a few seconds, and the resulting treatment plan is superior to that obtained using evenly distributed angles. Significance: A fast and automated personalized beam angle selection approach is been proposed for IMRT BAO.

physics.med-ph↗