Searcharxiv⌕ Search

arXiv subjects

Yuqing Wu

Publications and source records attributed to Yuqing Wu.

16 recordsLinked to original sources
arXiv2026

High-dimensional inference for the $\gamma$-ray sky with differentiable programming

We motivate the use of differentiable probabilistic programming techniques in order to account for the large model-space inherent to astrophysical $\gamma$-ray analyses. Targeting the longstanding Galactic Center $\gamma$-ray Excess (GCE) puzzle, we construct differentiable forward model and likelihood that make liberal use of GPU acceleration and vectorization in order to simultaneously account for a continuum of possible spatial morphologies consistent with the GCE emission in a fully probabilistic manner. Our setup allows for efficient inference over the large model space using variational methods. Beyond application to $\gamma$-ray data, a goal of this work is to showcase how differentiable probabilistic programming can be used as a tool to enable flexible analyses of astrophysical datasets.

astro-ph.HE↗
arXiv2026

TRIZ-RAGNER: A Retrieval-Augmented Large Language Model for TRIZ-Aware Named Entity Recognition in Patent-Based Contradiction Mining

TRIZ-based contradiction mining is a fundamental task in patent analysis and systematic innovation, as it enables the identification of improving and worsening technical parameters that drive inventive problem solving. However, existing approaches largely rely on rule-based systems or traditional machine learning models, which struggle with semantic ambiguity, domain dependency, and limited generalization when processing complex patent language. Recently, large language models (LLMs) have shown strong semantic understanding capabilities, yet their direct application to TRIZ parameter extraction remains challenging due to hallucination and insufficient grounding in structured TRIZ knowledge. To address these limitations, this paper proposes TRIZ-RAGNER, a retrieval-augmented large language model framework for TRIZ-aware named entity recognition in patent-based contradiction mining. TRIZ-RAGNER reformulates contradiction mining as a semantic-level NER task and integrates dense retrieval over a TRIZ knowledge base, cross-encoder reranking for context refinement, and structured LLM prompting to extract improving and worsening parameters from patent sentences. By injecting domain-specific TRIZ knowledge into the LLM reasoning process, the proposed framework effectively reduces semantic noise and improves extraction consistency. Experiments on the PaTRIZ dataset demonstrate that TRIZ-RAGNER consistently outperforms traditional sequence labeling models and LLM-based baselines. The proposed framework achieves a precision of 85.6%, a recall of 82.9%, and an F1-score of 84.2% in TRIZ contradiction pair identification. Compared with the strongest baseline using prompt-enhanced GPT, TRIZ-RAGNER yields an absolute F1-score improvement of 7.3 percentage points, confirming the effectiveness of retrieval-augmented TRIZ knowledge grounding for robust and accurate patent-based contradiction mining.

cs.CL↗
arXiv2025

AMAS: Adaptively Determining Communication Topology for LLM-based Multi-Agent System

Although large language models (LLMs) have revolutionized natural language processing capabilities, their practical implementation as autonomous multi-agent systems (MAS) for industrial problem-solving encounters persistent barriers. Conventional MAS architectures are fundamentally restricted by inflexible, hand-crafted graph topologies that lack contextual responsiveness, resulting in diminished efficacy across varied academic and commercial workloads. To surmount these constraints, we introduce AMAS, a paradigm-shifting framework that redefines LLM-based MAS through a novel dynamic graph designer. This component autonomously identifies task-specific optimal graph configurations via lightweight LLM adaptation, eliminating the reliance on monolithic, universally applied structural templates. Instead, AMAS exploits the intrinsic properties of individual inputs to intelligently direct query trajectories through task-optimized agent pathways. Rigorous validation across question answering, mathematical deduction, and code generation benchmarks confirms that AMAS systematically exceeds state-of-the-art single-agent and multi-agent approaches across diverse LLM architectures. Our investigation establishes that context-sensitive structural adaptability constitutes a foundational requirement for high-performance LLM MAS deployments.

cs.CL↗
arXiv2025

An automatic patent literature retrieval system based on LLM-RAG

With the acceleration of technological innovation efficient retrieval and classification of patent literature have become essential for intellectual property management and enterprise RD Traditional keyword and rulebased retrieval methods often fail to address complex query intents or capture semantic associations across technical domains resulting in incomplete and lowrelevance results This study presents an automated patent retrieval framework integrating Large Language Models LLMs with RetrievalAugmented Generation RAG technology The system comprises three components: 1) a preprocessing module for patent data standardization, 2) a highefficiency vector retrieval engine leveraging LLMgenerated embeddings, and 3) a RAGenhanced query module that combines external document retrieval with contextaware response generation Evaluations were conducted on the Google Patents dataset 20062024 containing millions of global patent records with metadata such as filing date domain and status The proposed gpt35turbo0125RAG configuration achieved 805 semantic matching accuracy and 92.1% recall surpassing baseline LLM methods by 28 percentage points The framework also demonstrated strong generalization in crossdomain classification and semantic clustering tasks These results validate the effectiveness of LLMRAG integration for intelligent patent retrieval providing a foundation for nextgeneration AIdriven intellectual property analysis platforms

cs.IR↗
arXiv2025

DynaSwarm: Dynamically Graph Structure Selection for LLM-based Multi-agent System

Current multi-agent systems (MAS) frameworks often rely on manually designed and static collaboration graph structures, limiting adaptability and performance. To address these limitations, we propose DynaSwarm, a dynamic framework that enhances LLM-based MAS through two key innovations: (1) an actor-critic reinforcement learning (A2C) mechanism to optimize graph structures with improved stability over prior RL methods, and (2) a dynamic graph selector that adaptively chooses the optimal graph structure for each input sample via parameter-efficient LLM fine-tuning. DynaSwarm eliminates the need for rigid, one-fits-all graph architectures, instead leveraging sample-specific idiosyncrasies to dynamically route queries through specialized agent networks. (c) We propose to fine-tune the demonstration retriever to fully exploit the power of in-context learning (ICL). Extensive experiments on question answering, mathematical reasoning, and coding tasks demonstrate that DynaSwarm consistently outperforms state-of-the-art single-agent and MAS baselines across multiple LLM backbones. Our findings highlight the importance of sample-aware structural flexibility in LLM MAS designs.

cs.LG↗
arXiv2025

Multi-Scale Target-Aware Representation Learning for Fundus Image Enhancement

High-quality fundus images provide essential anatomical information for clinical screening and ophthalmic disease diagnosis. Yet, due to hardware limitations, operational variability, and patient compliance, fundus images often suffer from low resolution and signal-to-noise ratio. Recent years have witnessed promising progress in fundus image enhancement. However, existing works usually focus on restoring structural details or global characteristics of fundus images, lacking a unified image enhancement framework to recover comprehensive multi-scale information. Moreover, few methods pinpoint the target of image enhancement, e.g., lesions, which is crucial for medical image-based diagnosis. To address these challenges, we propose a multi-scale target-aware representation learning framework (MTRL-FIE) for efficient fundus image enhancement. Specifically, we propose a multi-scale feature encoder (MFE) that employs wavelet decomposition to embed both low-frequency structural information and high-frequency details. Next, we design a structure-preserving hierarchical decoder (SHD) to fuse multi-scale feature embeddings for real fundus image restoration. SHD integrates hierarchical fusion and group attention mechanisms to achieve adaptive feature fusion while retaining local structural smoothness. Meanwhile, a target-aware feature aggregation (TFA) module is used to enhance pathological regions and reduce artifacts. Experimental results on multiple fundus image datasets demonstrate the effectiveness and generalizability of MTRL-FIE for fundus image enhancement. Compared to state-of-the-art methods, MTRL-FIE achieves superior enhancement performance with a more lightweight architecture. Furthermore, our approach generalizes to other ophthalmic image processing tasks without supervised fine-tuning, highlighting its potential for clinical applications.

eess.IV↗
arXiv2024

Robust Detection of LLM-Generated Text: A Comparative Analysis

The ability of large language models to generate complex texts allows them to be widely integrated into many aspects of life, and their output can quickly fill all network resources. As the impact of LLMs grows, it becomes increasingly important to develop powerful detectors for the generated text. This detector is essential to prevent the potential misuse of these technologies and to protect areas such as social media from the negative effects of false content generated by LLMS. The main goal of LLM-generated text detection is to determine whether text is generated by an LLM, which is a basic binary classification task. In our work, we mainly use three different classification methods based on open source datasets: traditional machine learning techniques such as logistic regression, k-means clustering, Gaussian Naive Bayes, support vector machines, and methods based on converters such as BERT, and finally algorithms that use LLMs to detect LLM-generated text. We focus on model generalization, potential adversarial attacks, and accuracy of model evaluation. Finally, the possible research direction in the future is proposed, and the current experimental results are summarized.

cs.CL↗
arXiv2024

Combining Induction and Transduction for Abstract Reasoning

When learning an input-output mapping from very few examples, is it better to first infer a latent function that explains the examples, or is it better to directly predict new test outputs, e.g. using a neural network? We study this question on ARC by training neural models for induction (inferring latent functions) and transduction (directly predicting the test output for a given test input). We train on synthetically generated variations of Python programs that solve ARC training tasks. We find inductive and transductive models solve different kinds of test problems, despite having the same training problems and sharing the same neural architecture: Inductive program synthesis excels at precise computations, and at composing multiple concepts, while transduction succeeds on fuzzier perceptual concepts. Ensembling them approaches human-level performance on ARC.

cs.LG↗
arXiv2023

Measuring the cosmological 21-cm dipole with 21-cm global experiments

A measurement of the 21-cm global signal would be a revealing probe of the Dark Ages, the era of first star formation, and the Epoch of Reionization. It has remained elusive owing to bright galactic and extra-galactic foreground contaminants, coupled with instrumental noise, ionospheric effects, and beam chromaticity. The simultaneous detection of a consistent 21-cm dipole signal alongside the 21-cm global signal would provide confidence in a claimed detection. We use simulated data to investigate the possibility of using drift-scan dipole antenna experiments to achieve a detection of both monopole and dipole. We find that at least two antennae located at different latitudes are required to localise the dipole. In the absence of foregrounds, a total integration time of $\sim 10^4$ hours is required to detect the dipole. With contamination by simple foregrounds, we find that the integration time required increases to $\sim 10^5$ hours. We show that the extraction of the 21-cm dipole from more realistic foregrounds requires a more sophisticated foreground modelling approach. Finally, we motivate a global network of dipole antennae that could reasonably detect the dipole in $\sim 10^3$ hours of integration time.

astro-ph.CO↗
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↗
arXiv2018

Comparing Downward Fragments of the Relational Calculus with Transitive Closure on Trees

Motivated by the continuing interest in the tree data model, we study the expressive power of downward navigational query languages on trees and chains. Basic navigational queries are built from the identity relation and edge relations using composition and union. We study the effects on relative expressiveness when we add transitive closure, projections, coprojections, intersection, and difference; this for boolean queries and path queries on labeled and unlabeled structures. In all cases, we present the complete Hasse diagram. In particular, we establish, for each query language fragment that we study on trees, whether it is closed under difference and intersection.

cs.DB↗
arXiv2017

Defect Line Terahertz Quantum Cascade Laser

We present terahertz quantum cascade lasers operating at a defect mode of a photonic crystal bandgap. This class of devices exhibits single mode emission and low threshold current compared to standard metal-metal lasers. The mode selectivity is an intrinsic property of the chosen fabrication design. The lower lasing threshold effect, already reported in photonic crystal quantum cascade lasers, is further enhanced in the ultra-flat-dispersion defect line. The presented results pave the way for integrated circuitry operating in the terahertz regime and have important applications in the field of quantum cascade lasers, spectroscopy and microcavity lasers.

physics.optics↗
arXiv2015

Structural characterizations of the navigational expressiveness of relation algebras on a tree

Given a document D in the form of an unordered node-labeled tree, we study the expressiveness on D of various basic fragments of XPath, the core navigational language on XML documents. Working from the perspective of these languages as fragments of Tarski's relation algebra, we give characterizations, in terms of the structure of D, for when a binary relation on its nodes is definable by an expression in these algebras. Since each pair of nodes in such a relation represents a unique path in D, our results therefore capture the sets of paths in D definable in each of the fragments. We refer to this perspective on language semantics as the "global view." In contrast with this global view, there is also a "local view" where one is interested in the nodes to which one can navigate starting from a particular node in the document. In this view, we characterize when a set of nodes in D can be defined as the result of applying an expression to a given node of D. All these definability results, both in the global and the local view, are obtained by using a robust two-step methodology, which consists of first characterizing when two nodes cannot be distinguished by an expression in the respective fragments of XPath, and then bootstrapping these characterizations to the desired results.

cs.DB↗
arXiv2014

Relative Expressive Power of Navigational Querying on Graphs

Motivated by both established and new applications, we study navigational query languages for graphs (binary relations). The simplest language has only the two operators union and composition, together with the identity relation. We make more powerful languages by adding any of the following operators: intersection; set difference; projection; coprojection; converse; and the diversity relation. All these operators map binary relations to binary relations. We compare the expressive power of all resulting languages. We do this not only for general path queries (queries where the result may be any binary relation) but also for boolean or yes/no queries (expressed by the nonemptiness of an expression). For both cases, we present the complete Hasse diagram of relative expressiveness. In particular the Hasse diagram for boolean queries contains some nontrivial separations and a few surprising collapses.

cs.DB↗
arXiv2013

External memory bisimulation reduction of big graphs

In this paper, we present, to our knowledge, the first known I/O efficient solutions for computing the k-bisimulation partition of a massive directed graph, and performing maintenance of such a partition upon updates to the underlying graph. Ubiquitous in the theory and application of graph data, bisimulation is a robust notion of node equivalence which intuitively groups together nodes in a graph which share fundamental structural features. k-bisimulation is the standard variant of bisimulation where the topological features of nodes are only considered within a local neighborhood of radius $k\geqslant 0$. The I/O cost of our partition construction algorithm is bounded by $O(k\cdot \mathit{sort}(|\et|) + k\cdot scan(|\nt|) + \mathit{sort}(|\nt|))$, while our maintenance algorithms are bounded by $O(k\cdot \mathit{sort}(|\et|) + k\cdot \mathit{sort}(|\nt|))$. The space complexity bounds are $O(|\nt|+|\et|)$ and $O(k\cdot|\nt|+k\cdot|\et|)$, resp. Here, $|\et|$ and $|\nt|$ are the number of disk pages occupied by the input graph's edge set and node set, resp., and $\mathit{sort}(n)$ and $\mathit{scan}(n)$ are the cost of sorting and scanning, resp., a file occupying $n$ pages in external memory. Empirical analysis on a variety of massive real-world and synthetic graph datasets shows that our algorithms perform efficiently in practice, scaling gracefully as graphs grow in size.

cs.DB↗