arXiv subjects

Xiang Zou

Publications and source records attributed to Xiang Zou.

14 recordsLinked to original sources

arXiv2026

SPDCN: Strip-based Deformable Convolutional Network for Steel Surface Defect Segmentation

Steel surface defect segmentation is critical for industrial quality inspection, yet existing methods struggle with elongated, anisotropic defects such as cracks and scratches due to the isotropic receptive fields of standard convolutions and rigid sampling grids that cannot adapt to irregular defect boundaries. To address these limitations, we propose Strip-based Predictor for Deformable Convolutional Networks (SPDCN) with two key innovations. The \textbf{Fuzzy-enhanced Multi-scale Context Module (FMCM)} employs group-wise multi-branch convolutions with an intuitionistic fuzzy channel attention mechanism to adaptively capture multi-scale contextual information across varying defect sizes. The \textbf{Adaptive Direction-Aware Deformable Convolution (ADADC)} replaces the conventional offset predictor with decoupled horizontal and vertical strip convolutions, enabling the deformable sampling grid to anisotropically align with the principal orientation of elongated defects. Extensive experiments on public steel surface defect benchmarks demonstrate that SPDCN consistently outperforms state-of-the-art methods, achieving 89.60\% mIoU on NEU-Seg with only 3.54M parameters. The source code is publicly available at https://github.com/DWlzm .

cs.CV↗

arXiv2026

Stellar discs and intermediate-mass black holes in galactic nuclei I. Fragmenting the disc in an isotropic stellar potential

The origin of the complex orbital structure of young massive stars at the Galactic centre remains an open question. If these stars formed in a single episode from a gaseous accretion disc, they may initially have constituted a single, coherently rotating stellar disc. We investigate whether perturbations from an unseen intermediate-mass black hole (IMBH) could fragment and/or disrupt such a disc into the multiple orbital components observed today. First, we derive a theoretical criterion for when and where the IMBH's torque overcomes the disc's self-torque and tears it apart. We then test this picture with direct $N$-body simulations of a stellar disc interacting with an inclined IMBH around a central supermassive black hole. We find that the outcome depends strongly on the IMBH's orbit and mass. A prograde IMBH rapidly aligns with the stellar disc, while a massive retrograde IMBH ($m_{\bullet} \simeq 0.67\,M_{\rm d}$) anti-aligns relative to the radially overlapping stars and efficiently fragments the original disc into three components in angular-momentum space: an inner disc, a misaligned overlapping region, and an unperturbed outer disc. The IMBH also excites eccentricities in the overlapping region, driving stars away from initially circular orbits. These features emerge for an IMBH mass of $2000\,{\rm M}_{\odot}$ and a disc mass of $3000\,{\rm M}_{\odot}$ within 10--20 Myr, a timescale comparable to the age of the young Galactic centre stellar population, and provide a plausible explanation for the observed multiple orbital planes, warped geometry, and broad eccentricity distribution.

astro-ph.GA↗

arXiv2026

Beyond Instance-Level Self-Supervision in 3D Multi-Modal Medical Imaging

Self-supervised pre-training methods in medical imaging typically treat each individual as an isolated instance, learning representations through augmentation-based objectives or masked reconstruction. They often do not adequately capitalize on a key characteristic of physiological features: anatomical structures maintain consistent spatial relationships across individuals (instances), such as the thalamus being medial to the basal ganglia, regardless of variations in brain size, shape, or pathology. We propose leveraging this cross-instance topological consistency as a supervisory signal. The challenge arises from the inherent variability in medical imaging, which can differ significantly across instances and modalities. To tackle this, we focus on two alignment regimes. (i) Intra-instance: with pixel-level correspondences available, a cross-modal triplet objective explicitly preserves local neighborhood topology. (ii) Inter-instance: without such supervision, we derive pseudo-correspondences to control partial neighborhood alignment and prevent topology collapse across modalities. We validate our approach across 7 downstream multi-modal tasks, achieving average improvements of 1.1% and 5.94% in segmentation and classification tasks, respectively, and demonstrating significantly better robustness when modalities are missing at test time.

cs.CV↗

arXiv2025

Evaluating Large Language Models in Scientific Discovery

Large language models (LLMs) are increasingly applied to scientific research, yet prevailing science benchmarks probe decontextualized knowledge and overlook the iterative reasoning, hypothesis generation, and observation interpretation that drive scientific discovery. We introduce a scenario-grounded benchmark that evaluates LLMs across biology, chemistry, materials, and physics, where domain experts define research projects of genuine interest and decompose them into modular research scenarios from which vetted questions are sampled. The framework assesses models at two levels: (i) question-level accuracy on scenario-tied items and (ii) project-level performance, where models must propose testable hypotheses, design simulations or experiments, and interpret results. Applying this two-phase scientific discovery evaluation (SDE) framework to state-of-the-art LLMs reveals a consistent performance gap relative to general science benchmarks, diminishing return of scaling up model sizes and reasoning, and systematic weaknesses shared across top-tier models from different providers. Large performance variation in research scenarios leads to changing choices of the best performing model on scientific discovery projects evaluated, suggesting all current LLMs are distant to general scientific "superintelligence". Nevertheless, LLMs already demonstrate promise in a great variety of scientific discovery projects, including cases where constituent scenario scores are low, highlighting the role of guided exploration and serendipity in discovery. This SDE framework offers a reproducible benchmark for discovery-relevant evaluation of LLMs and charts practical paths to advance their development toward scientific discovery.

cs.AI↗

arXiv2025

T-GRAG: A Dynamic GraphRAG Framework for Resolving Temporal Conflicts and Redundancy in Knowledge Retrieval

Large language models (LLMs) have demonstrated strong performance in natural language generation but remain limited in knowle- dge-intensive tasks due to outdated or incomplete internal knowledge. Retrieval-Augmented Generation (RAG) addresses this by incorporating external retrieval, with GraphRAG further enhancing performance through structured knowledge graphs and multi-hop reasoning. However, existing GraphRAG methods largely ignore the temporal dynamics of knowledge, leading to issues such as temporal ambiguity, time-insensitive retrieval, and semantic redundancy. To overcome these limitations, we propose Temporal GraphRAG (T-GRAG), a dynamic, temporally-aware RAG framework that models the evolution of knowledge over time. T-GRAG consists of five key components: (1) a Temporal Knowledge Graph Generator that creates time-stamped, evolving graph structures; (2) a Temporal Query Decomposition mechanism that breaks complex temporal queries into manageable sub-queries; (3) a Three-layer Interactive Retriever that progressively filters and refines retrieval across temporal subgraphs; (4) a Source Text Extractor to mitigate noise; and (5) a LLM-based Generator that synthesizes contextually and temporally accurate responses. We also introduce Time-LongQA, a novel benchmark dataset based on real-world corporate annual reports, designed to test temporal reasoning across evolving knowledge. Extensive experiments show that T-GRAG significantly outperforms prior RAG and GraphRAG baselines in both retrieval accuracy and response relevance under temporal constraints, highlighting the necessity of modeling knowledge evolution for robust long-text question answering. Our code is publicly available on the T-GRAG

cs.AI↗

arXiv2025

Graph Counselor: Adaptive Graph Exploration via Multi-Agent Synergy to Enhance LLM Reasoning

Graph Retrieval Augmented Generation (GraphRAG) effectively enhances external knowledge integration capabilities by explicitly modeling knowledge relationships, thereby improving the factual accuracy and generation quality of Large Language Models (LLMs) in specialized domains. However, existing methods suffer from two inherent limitations: 1) Inefficient Information Aggregation: They rely on a single agent and fixed iterative patterns, making it difficult to adaptively capture multi-level textual, structural, and degree information within graph data. 2) Rigid Reasoning Mechanism: They employ preset reasoning schemes, which cannot dynamically adjust reasoning depth nor achieve precise semantic correction. To overcome these limitations, we propose Graph Counselor, an GraphRAG method based on multi-agent collaboration. This method uses the Adaptive Graph Information Extraction Module (AGIEM), where Planning, Thought, and Execution Agents work together to precisely model complex graph structures and dynamically adjust information extraction strategies, addressing the challenges of multi-level dependency modeling and adaptive reasoning depth. Additionally, the Self-Reflection with Multiple Perspectives (SR) module improves the accuracy and semantic consistency of reasoning results through self-reflection and backward reasoning mechanisms. Experiments demonstrate that Graph Counselor outperforms existing methods in multiple graph reasoning tasks, exhibiting higher reasoning accuracy and generalization ability. Our code is available at https://github.com/gjq100/Graph-Counselor.git.

cs.AI↗

arXiv2025

Algebraic Topology Principles behind Topological Quantum Error Correction

Quantum error correction (QEC) is crucial for realizing scalable quantum technologies, and topological quantum error correction (TQEC) has emerged as the most experimentally advanced paradigm of QEC. Existing homological and topological code constructions, however, are largely confined to orientable two-manifolds with simple boundary conditions. In this work, we develop a unified algebraic-topological framework for TQEC based on homology, cohomology, and intersection theory, which characterizes exactly when an arbitrary-dimensional manifold (with or without boundary) can serve as a quantum memory, thereby extending the standard 2D homological-code picture to arbitrary dimension and to manifolds with boundary via Poincar\'e-Lefschetz duality. Building on this classification, we introduce concrete code families that exploit nontrivial topology beyond the planar and toric settings. These include ``3-torus code'' and higher-dimensional ``volume codes'' on compact manifolds with mixed $X$- and $Z$-type boundaries. We further give a topological construction of qudit TQEC codes on general two-dimensional cell complexes using group presentation complexes, which unifies and extends several known quantum LDPC and homological-product-like constructions within a single geometric language. Finally, we combine the theoretical framework with numerical simulations to demonstrate that changing only the global topology can yield improved logical performance at fixed entanglement resources. Taken together, our results provide a systematic set of topological design principles for constructing and analyzing TQEC codes across dimensions and boundaries, and they open new avenues for topology-aware fault-tolerant quantum architectures.

quant-ph↗

arXiv2024

A square cross-section FOV rotational CL (SC-CL) and its analytical reconstruction method

Rotational computed laminography (CL) has broad application potential in three-dimensional imaging of plate-like objects, as it only needs x-ray to pass through the tested object in the thickness direction during the imaging process. In this study, a square cross-section FOV rotational CL (SC-CL) was proposed. Then, the FDK-type analytical reconstruction algorithm applicable to the SC-CL was derived. On this basis, the proposed method was validated through numerical experiments.

eess.IV↗

arXiv2023

Blockchain-Envisioned UAV-Aided Disaster Relief Networks: Challenges and Solutions

Natural or man-made disasters pose significant challenges for delivering critical relief to affected populations due to disruptions in critical infrastructures and logistics networks. Unmanned aerial vehicles (UAVs)-aided disaster relief networks (UDRNs) leverage UAVs to assist existing ground relief networks by swiftly assessing affected areas and timely delivering lifesaving supplies. To meet the growing demands for collaborative, trust-free, and transparent UDRN services, blockchain-based UDRNs emerge as a promising approach through immutable ledgers and distributed smart contracts. However, several efficiency and security challenges hinder the deployment of blockchain-based UDRNs, including the lack of cooperation between smart contracts, lack of dynamic audit for smart contract vulnerabilities, and low forensics robustness against transaction malleability attacks. Towards efficient and secure blockchain-based UDRNs, this paper presents potential solutions: (i) a series of collaborative smart contracts for coordinated relief management, (ii) a dynamic contract audit mechanism to prevent known/unknown contract vulnerabilities; and (iii) a robust transaction forensics strategy with on/off-chain cooperation to resist transaction malleability attacks. Our prototype implementation and experimental results demonstrate the feasibility and effectiveness of our approach. Lastly, we outline key open research issues crucial to advancing this emerging field.

cs.CR↗

arXiv2022

An LLVM-based C++ Compiler Toolchain for Variational Hybrid Quantum-Classical Algorithms and Quantum Accelerators

Variational algorithms are a representative class of quantum computing workloads that combine quantum and classical computing. This paper presents an LLVM-based C++ compiler toolchain to efficiently execute variational hybrid quantum-classical algorithms on a computational system in which the quantum device acts as an accelerator. We introduce a set of extensions to the C++ language for programming these algorithms. We define a novel Executable and Linking Format (ELF) for Quantum and create a quantum device compiler component in the LLVM framework to compile the quantum part of the C++ source and reuse the host compiler in the LLVM framework to compile the classical computing part of the C++ source. A variational algorithm runs a quantum circuit repeatedly, each time with different gate parameters. We add to the quantum runtime the capability to execute dynamically a quantum circuit with different parameters. Thus, programmers can call quantum routines the same way as classical routines. With these capabilities, a variational hybrid quantum-classical algorithm can be specified in a single-source code and only needs to be compiled once for all iterations. The single compilation significantly reduces the execution latency of variational algorithms. We evaluate the framework's performance by running quantum circuits that prepare Thermofield Double (TFD) states, a quantum-classical variational algorithm.

quant-ph↗

arXiv2021

Speculative Interference Attacks: Breaking Invisible Speculation Schemes

Recent security vulnerabilities that target speculative execution (e.g., Spectre) present a significant challenge for processor design. The highly publicized vulnerability uses speculative execution to learn victim secrets by changing cache state. As a result, recent computer architecture research has focused on invisible speculation mechanisms that attempt to block changes in cache state due to speculative execution. Prior work has shown significant success in preventing Spectre and other vulnerabilities at modest performance costs. In this paper, we introduce speculative interference attacks, which show that prior invisible speculation mechanisms do not fully block these speculation-based attacks. We make two key observations. First, misspeculated younger instructions can change the timing of older, bound-to-retire instructions, including memory operations. Second, changing the timing of a memory operation can change the order of that memory operation relative to other memory operations, resulting in persistent changes to the cache state. Using these observations, we demonstrate (among other attack variants) that secret information accessed by mis-speculated instructions can change the order of bound-to-retire loads. Load timing changes can therefore leave secret-dependent changes in the cache, even in the presence of invisible speculation mechanisms. We show that this problem is not easy to fix: Speculative interference converts timing changes to persistent cache-state changes, and timing is typically ignored by many cache-based defenses. We develop a framework to understand the attack and demonstrate concrete proof-of-concept attacks against invisible speculation mechanisms. We provide security definitions sufficient to block speculative interference attacks; describe a simple defense mechanism with a high performance cost; and discuss how future research can improve its performance.

cs.AR↗

arXiv2020

Enhancing a Near-Term Quantum Accelerator's Instruction Set Architecture for Materials Science Applications

Quantum computers with tens to hundreds of noisy qubits are being developed today. To be useful for real-world applications, we believe that these near-term systems cannot simply be scaled-down non-error-corrected versions of future fault-tolerant large-scale quantum computers. These near-term systems require specific architecture and design attributes to realize their full potential. To efficiently execute an algorithm, the quantum coprocessor must be designed to scale with respect to qubit number and to maximize useful computation within the qubits' decoherence bounds. In this work, we employ an application-system-qubit co-design methodology to architect a near-term quantum coprocessor. To support algorithms from the real-world application area of simulating the quantum dynamics of a material system, we design a (parameterized) arbitrary single-qubit rotation instruction and a two-qubit entangling controlled-Z instruction. We introduce dynamic gate set and paging mechanisms to implement the instructions. To evaluate the functionality and performance of these two instructions, we implement a two-qubit version of an algorithm to study a disorder-induced metal-insulator transition and run 60 random instances of it, each of which realizes one disorder configuration and contains 40 two-qubit instructions (or gates) and 104 single-qubit instructions. We observe the expected quantum dynamics of the time-evolution of this system.

quant-ph↗

arXiv2020

A Thermodynamic Model for Active Ion Transport

Active ion transport is very critical for living cells to maintain and regular internal and external environment. It is known that adenosine 5'-triphosphate (ATP) is a general energy source that applies bond energy for pumps to overcome ion concentration gradient. In this study, we introduce a novel thermodynamic model for active ion transport, which allows the pump to act as a 'Maxwell's demon', and also conforms to the second law of thermodynamics. Transport against the gradient is caused by the thermodynamic fluctuation with information recording in ATP stream. Besides, exhaustive experiments about Na-K ATPase performed in the red cell ghost system are reviewed to verify this model, and all the results can support the solvable theory about the working mechanism of demon pumps. Our findings indicate the possible role of ATP as an information carrier but not energy currency. The high-energy phosphate bonds can improve the efficiency of information recording in relevance to ion transport and may mainly convert to heat.

q-bio.SC↗

arXiv2019

The Intrinsic Properties of Brain Based on the Network Structure

Objective: Brain is a fantastic organ that helps creature adapting to the environment. Network is the most essential structure of brain, but the capability of a simple network is still not very clear. In this study, we try to expound some brain functions only by the network property. Methods: Every network can be equivalent to a simplified network, which is expressed by an equation set. The dynamic of the equation set can be described by some basic equations, which is based on the mathematical derivation. Results (1) In a closed network, the stability is based on the excitatory/inhibitory synapse proportion. Spike probabilities in the assembly can meet the solution of a nonlinear equation set. (2) Network activity can spontaneously evolve into a certain distribution under different stimulation, which is closely related to decision making. (3) Short memory can be formed by coupling of network assemblies. Conclusion: The essential property of a network may contribute to some important brain functions.

q-bio.NC↗