arXiv subjects

Shan Zhang

Publications and source records attributed to Shan Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Diagram-MMU: A Multi-Modal Benchmark for Scientific Diagrams

Multimodal Large Language Models (MLLMs) have been growing the capability for scientific writing and collaboration. For example, OpenAI Prism is a free workspace for scientific writing and collaboration. One important feature in Prism is turning scientific diagrams directly into LaTeX TikZ code. In this paper, we build a benchmark, Diagram-MMU, a multi-modal benchmark designed to assess MLLMs' ability for scientific diagram parsing and understanding. Diagram-MMU features 3.7k curated diagrams and 18.3k human-validated questions across six domains. It evaluates MLLMs on three tasks common in vibe writing workspaces: diagram-to-code parsing, diagram-to-code editing, and diagram question answering, alongside agentic settings per task. The evaluation of 12 MLLMs reveals that diagram-to-code tasks are more challenging than diagram question answering: models can reason well over diagrams but struggle to parse and edit them, underscoring the need for methods to enhance MLLMs' capability in diagram-to-code generation. Under agentic settings, most models improve parsing and editing performance but degrade on question answering, while Claude-4.6 Opus consistently improves across all three tasks. Project Page: https://vi-ocean.github.io/projects/diagram-mmu.

cs.CV↗

arXiv2026

Bi-S network origin of cation-disorder stability and dispersive band edges in AgBiS2

Cation-disordered AgBiS2 is a promising lead-free optoelectronic material, but both its ordered structure and the microscopic origin of its favorable electronic properties remain debated. Theory has proposed a mixed-coordination tendency with tetrahedral AgS4 and octahedral BiS6 units, whereas experiments mainly report octahedrally coordinated ordered and cation-disordered phases, together with local cation off-centering. Here, we combine a machine-learning interatomic potential with a deep-learning Hamiltonian to resolve the coupled structural and electronic evolution of AgBiS2 at large length scales. We identify the three-dimensional Bi-S network as the central structural motif governing both disorder stability and band-edge electronic states. At weak disorder, Ag/Bi exchange competes with the off-centering tendency of the Ag sublattice, producing strongly distorted local environments and convoluted diffraction signatures that hinder the identification of the ordered phase. With increasing disorder, BiS6-like units connect into a continuous Bi-S network, which stabilizes the rocksalt-like disordered phase. Despite strong cation disorder, AgBiS2 retains clear semiconductor-like band dispersion and develops a direct band gap. The connected Bi:p-S:p states supported by the Bi-S network preserve a dispersive conduction-band edge and a small electron effective mass. In contrast, mobile Ag disrupts the long-range periodicity of Ag-S bonding, leading to strongly localized valence states. These results clarify the structural controversy in ordered AgBiS2 and establish a unified physical picture of disorder stability and optoelectronic response in nonisovalent semiconductor alloys.

cond-mat.mtrl-sci↗

arXiv2026

FiSeR: Fine-Grained Source Representations for Cross-Domain AI Image Detection

Real-world synthetic image detectors often generalize poorly under domain shift despite strong in-domain performance. Using unsupervised UMAP projections, we find that natural and synthetic features remain partially separable on unseen datasets, yet performance still drops, suggesting that the classification head overfits to training-domain artifacts. Therefore, the key is to learn more transferable representations so that the decision criterion is more stable and robust to domain shifts. Based on the structural fact that synthetic images are produced by diverse generators, we propose a hierarchical contrastive learning framework that improves the separability between natural and synthetic images while preserving generator identity information. It jointly optimizes (i) a coarse contrastive objective between natural and synthetic images and (ii) a fine contrastive objective among synthetic images using generator identities. Trained on WildFake, our method achieves an average AUROC gain of +10.22 on cross-domain evaluation over Chameleon, AIGIBench, Community Forensics, and GenImage under the same settings as the strong baseline DIRE. For few-shot adaptation, we freeze the backbone and fit an SVM head on 10 labeled samples per class, improving AUROC by +10.64 on AIGIBench and +17.41 on Chameleon, averaged over 12 widely used detectors. Our code is publicly available at: https://github.com/heyongxin233/FiSeR.

cs.CV↗

arXiv2026

Analyzing Middle School Students' Dialogue and Behaviors during Collaborative AI Chatbot Development Using Ordered Network Analysis

As Artificial Intelligence (AI) education has become a key component of K-12 curricula, activities such as designing and developing conversational agents are increasingly used as instructional practice. Prior work has primarily examined these activities by focusing on students' learning outcomes or the quality of final AI artifacts, offering limited insight into the collaborative processes through which learning unfolds during AI system development. Although the AIED community has a long history of studying collaborative learning in STEM and Computing education, the emergence of AI learning environments in which students build AI systems presents new opportunities to understand how collaboration unfolds in AI education contexts. Grounded in these foundational works, the current study examines collaborative interaction among middle school students engaged in the design and development of an AI chatbot. Using Ordered Network Analysis of students' dialogue and development actions, we characterize how collaboration is organized over time and how interaction patterns relate to chatbot quality and AI knowledge outcomes. Results reveal that higher-quality chatbots are associated with more integrated sequences linking explanation, testing, and refinement. Interaction patterns involving articulated reasoning and repeated testing and revision in response to chatbot output were also associated with stronger AI knowledge outcomes. These findings provide a process-oriented account of collaborative AI chatbot development and extend AIED research on collaborative learning processes to AI education contexts.

cs.HC↗

arXiv2026

Small, Private Language Models as Teammates for Educational Assessment Design

Generative AI increasingly supports educational design tasks, e.g., through Large Language Models (LLMs), demonstrating the capability to design assessment questions that are aligned with pedagogical frameworks (e.g., Bloom's taxonomy). However, they often rely on subjective or limited evaluation methods; focus primarily on proprietary models; or rarely systematically examine generation, evaluation, or deployment constraints in real educational settings. Meanwhile, Small Language Models (SLMs) have emerged as local alternatives that better address privacy and resource limitations; yet their effectiveness for assessment tasks remains underexplored. To address this gap, we systematically compare LLMs and SLMs for assessment question design; evaluate generation quality across Bloom's taxonomy levels using reproducible, pedagogically grounded metrics; and further assess model-based judging against expert-informed evaluation by analyzing reliability and agreement patterns. Results show that SLMs achieve competitive performance across key pedagogically motivated quality dimensions while enabling local, privacy-sensitive deployment. However, model-based evaluations also exhibit systematic inconsistencies and bias relative to expert ratings. These findings provide evidence to posit language models as bounded assistants in assessment workflows; underscore the necessity of Human-in-the-Loop; and advance the automated educational question generation field by examining quality, reliability, and deployment-aware trade-offs.

cs.AI↗

arXiv2026

Interactive Evidence Maps for Visualizing and Understanding Systematic Reviews

Systematic reviews provide comprehensive syntheses of research fields. As a result, systematic reviews often emphasize synthesizing across the large bodies of literature rather than just describing the studies from which the conclusions were drawn. This risks an incomplete description of the sample - encouraging overgeneralization of the findings, obscuring connections between existing work, or overshadowing gaps in the literature. To address this challenge, we introduce interactive evidence maps; an accessible visualization tool that enables researchers to explore, filter, and analyze review data dynamically. Our approach leverages large language models to extract topic models that structure heterogeneous review data into an interactive, explorable knowledge map that supports deeper inspection beyond static tables and figures. We demonstrate the usefulness of interactive evidence maps using data from a published scoping review of pedagogical agents in K-12 education, and compare the results of the evidence map to those reported in the scoping review. Results show that interactive evidence maps complement traditional syntheses by enhancing transparency, supporting exploratory analysis, and revealing patterns and gaps that may not be easy to detect through narrative summaries alone.

cs.DL↗

arXiv2026

Presence versus absence of charging energies in PbTe quantum dots

Charging energy ($E_C$) is essential in quantum dot (QD) devices. Previous studies on PbTe QDs have reported both the presence and absence of $E_C$. To resolve this ambiguity, we vary the QD size, i.e. the cross-sectional area of PbTe nanowires, and track the evolution of $E_C$. For large crosssectional areas ($\sim$ 16000 nm$^2$), the PbTe QDs exhibit no measurable $E_C$, while quantized levels are well resolved. Decreasing this area successively to 5000, 1500, and 460 nm$^2$, $E_C$ becomes finite and increases to 80, 160, and 210 $\mu$eV, respectively. We further demonstrate the strong tunability of local gates, which can tune the PbTe device from the QD regime to the regime of ballistic transport. These results address concerns regarding the large dielectric constant of PbTe and provide key insights in engineering advanced PbTe quantum devices.

cond-mat.mes-hall↗

arXiv2026

How to Assess AI Literacy: Misalignment Between Self-Reported and Objective-Based Measures

The widespread adoption of Artificial Intelligence (AI) in K-12 education highlights the need for psychometrically-tested measures of teachers' AI literacy. Existing work has primarily relied on either self-report (SR) or objective-based (OB) assessments, with few studies aligning the two within a shared framework to compare perceived versus demonstrated competencies or examine how prior AI literacy experience shapes this relationship. This gap limits the scalability of learning analytics and the development of learner profile-driven instructional design. In this study, we developed and evaluated SR and OB measures of teacher AI literacy within the established framework of Concept, Use, Evaluate, and Ethics. Confirmatory factor analyses support construct validity with good reliability and acceptable fit. Results reveal a low correlation between SR and OB factors. Latent profile analysis identified six distinct profiles, including overestimation (SR > OB), underestimation (SR < OB), alignment (SR close to OB), and a unique low-SR/low-OB profile among teachers without AI literacy experience. Theoretically, this work extends existing AI literacy frameworks by validating SR and OB measures on shared dimensions. Practically, the instruments function as diagnostic tools for professional development, supporting AI-informed decisions (e.g., growth monitoring, needs profiling) and enabling scalable learning analytics interventions tailored to teacher subgroups.

cs.CY↗

arXiv2026

Let Me Try Again: Examining Replay Behavior by Tracing Students' Latent Problem-Solving Pathways

Prior research has shown that students' problem-solving pathways in game-based learning environments reflect their conceptual understanding, procedural knowledge, and flexibility. Replay behaviors, in particular, may indicate productive struggle or broader exploration, which in turn foster deeper learning. However, little is known about how these pathways unfold sequentially across problems or how the timing of replays and other problem-solving strategies relates to proximal and distal learning outcomes. This study addresses these gaps using Markov Chains and Hidden Markov Models (HMMs) on log data from 777 seventh graders playing the game-based learning platform of From Here to There!. Results show that within problem sequences, students often persisted in states or engaged in immediate replay after successful completions, while across problems, strong self-transitions indicated stable strategic pathways. Four latent states emerged from HMMs: Incomplete-dominant, Optimal-ending, Replay, and Mixed. Regression analyses revealed that engagement in replay-dominant and optimal-ending states predicted higher conceptual knowledge, flexibility, and performance compared with the Incomplete-dominant state. Immediate replay consistently supported learning outcomes, whereas delayed replay was weakly or negatively associated in relation to Non-Replay. These findings suggest that replay in digital learning is not uniformly beneficial but depends on timing, with immediate replay supporting flexibility and more productive exploration.

cs.CY↗

arXiv2026

MAESTRO: Multi-Agent Evaluation Suite for Testing, Reliability, and Observability

We present MAESTRO, an evaluation suite for the testing, reliability, and observability of LLM-based MAS. MAESTRO standardizes MAS configuration and execution through a unified interface, supports integrating both native and third-party MAS via a repository of examples and lightweight adapters, and exports framework-agnostic execution traces together with system-level signals (e.g., latency, cost, and failures). We instantiate MAESTRO with 12 representative MAS spanning popular agentic frameworks and interaction patterns, and conduct controlled experiments across repeated runs, backend models, and tool configurations. Our case studies show that MAS executions can be structurally stable yet temporally variable, leading to substantial run-to-run variance in performance and reliability. We further find that MAS architecture is the dominant driver of resource profiles, reproducibility, and cost-latency-accuracy trade-off, often outweighing changes in backend models or tool settings. Overall, MAESTRO enables systematic evaluation and provides empirical guidance for designing and optimizing agentic systems.

cs.NI↗

arXiv2025

Linking Thermal History to Shear Band Interaction and Macroscopic Ductility in Metallic Glasses

Shear band propagation and interaction are critical to the mechanical performance of metallic glasses and are strongly governed by thermal history, yet their microscopic mechanisms remain unclear. Here, using molecular dynamics simulations combined with a state-of-the-art annealing protocol, we systematically investigate these behaviors in a model metallic glass across effective quenching rates spanning six orders of magnitude. Through a double-notch model, we show that the normalized interaction distance relative to the single shear band width is significantly larger in slowly quenched samples than in rapidly quenched ones. Atomic-scale analysis reveals that rapidly quenched samples exhibit a high density of pre-existing soft regions, which trigger correlated shear transformation zones through local vortex fields, resulting in propagation path locking and weak inter-band coupling. In contrast, slowly quenched samples exhibit enhanced structural heterogeneity and a right-shifted activation energy spectrum, promoting a single large-scale vortex field ahead of the shear band front. This field facilitates long-range stress transmission and induces shear band deflection, convergence, and coalescence, a transition resembling a "shielding effect" in fracture mechanics, where vortex-mediated disturbances destabilize the advancing shear band front. Our findings establish a direct microscopic connection between glass stability and shear-band-mediated plasticity and suggest that regulating shear band interactions offers a promising route to enhance the room-temperature ductility of metallic glasses.

cond-mat.mtrl-sci↗

arXiv2025

Turing Pattern Engineering Enables Kinetically Ultrastable yet Ductile Metallic Glasses

Enhancing the kinetic stability of glasses often necessitates deepening thermodynamic stability, which typically compromises ductility due to increased structural rigidity. Decoupling these properties remains a critical challenge for functional applications. Here, we demonstrate that pattern engineering in metallic glasses (MGs) enables unprecedented kinetic ultrastability while retaining thermodynamic metastability and intrinsic plasticity. Through atomistic simulations guided by machine-learning interatomic potentials and replica-exchange molecular dynamics, we reveal that clustering oxygen contents, driven by reaction-diffusion-coupled pattern dynamics, act as localized pinning sites. These motifs drastically slow structural relaxation, yielding kinetic stability comparable to crystal-like ultrastable glasses while retaining an energetic as-cast state. Remarkably, the thermodynamically metastable state preserves heterogeneous atomic mobility, allowing strain delocalization under mechanical stress. By tailoring oxygen modulation via geometric patterning, we achieve an approximately 200 K increase in the onset temperature of the glass transition (Tonset) while maintaining fracture toughness akin to conventional MGs. This work establishes a paradigm of kinetic stabilization without thermodynamic compromise, offering a roadmap to additively manufacture bulk amorphous materials with combined hyperstability and plasticity.

cond-mat.mtrl-sci↗

arXiv2025

Hierarchical Process Reward Models are Symbolic Vision Learners

Symbolic computer vision represents diagrams through explicit logical rules and structured representations, enabling interpretable understanding in machine vision. This requires fundamentally different learning paradigms from pixel-based visual models. Symbolic visual learners parse diagrams into geometric primitives-points, lines, and shapes-whereas pixel-based learners operate on textures and colors. We propose a novel self-supervised symbolic auto-encoder that encodes diagrams into structured primitives and their interrelationships within the latent space, and decodes them through our executable engine to reconstruct the input diagrams. Central to this architecture is Symbolic Hierarchical Process Reward Modeling, which applies hierarchical step-level parsing rewards to enforce point-on-line, line-on-shape, and shape-on-relation consistency. Since vanilla reinforcement learning exhibits poor exploration in the policy space during diagram reconstruction; we thus introduce stabilization mechanisms to balance exploration and exploitation. We fine-tune our symbolic encoder on downstream tasks, developing a neuro-symbolic system that integrates the reasoning capabilities of neural networks with the interpretability of symbolic models through reasoning-grounded visual rewards. Evaluations across reconstruction, perception, and reasoning tasks demonstrate the effectiveness of our approach: achieving a 98.2% reduction in MSE for geometric diagram reconstruction, surpassing GPT-4o by 0.6% with a 7B model on chart reconstruction, and improving by +13% on the MathGlance perception benchmark, and by +3% on MathVerse and GeoQA reasoning benchmarks.

cs.CV↗

arXiv2025

Artemis: Structured Visual Reasoning for Perception Policy Learning

Recent reinforcement-learning frameworks for visual perception policy usually incorporate intermediate reasoning chains expressed in natural language. Empirical observations indicate that such purely linguistic intermediate reasoning often reduces performance on perception tasks. We argue that the core issue lies not in reasoning per se but in the form of reasoning: while these chains perform semantic reasoning in an unstructured linguistic space, \textbf{visual perception requires reasoning in a spatial and object-centric space}. In response, we introduce \textbf{Artemis}, a perception-policy learning method that performs structured visual reasoning, where each intermediate step is represented as a (label, bounding-box) pair capturing a verifiable visual state. This design enables explicit tracking of intermediate states, direct supervision for proposal quality, and avoids ambiguity introduced by language-based reasoning. Building upon verifiable and spatially grounded reasoning chains, Artemis provides a unified architecture for diverse perceptual tasks, without requiring the task-specific designs relied upon by prior perceptual policy models. Trained using grounding and detection sampeles in natural image domains, Artemis generalizes to counting and geometric perception tasks. At its core, a spatially grounded, object-centric chain rule provides a principled foundation for scalable and general perceptual policies.

cs.CV↗

arXiv2025

Agentic Learner with Grow-and-Refine Multimodal Semantic Memory

MLLMs exhibit strong reasoning on isolated queries, yet they operate de novo -- solving each problem independently and often repeating the same mistakes. Existing memory-augmented agents mainly store past trajectories for reuse. However, trajectory-based memory suffers from brevity bias, gradually losing essential domain knowledge. More critically, even in truly multimodal problem-solving settings, it records only a single-modality trace of past behavior, failing to preserve how visual attention and logical reasoning jointly contributed to the solution. This is fundamentally misaligned with human cognition: semantic memory is both multimodal and integrated, preserving visual and abstract knowledge through coordinated but distinct representational streams. We thus introduce ViLoMem, a dual-stream memory framework that constructs compact, schema-based memory. It separately encodes visual distraction patterns and logical reasoning errors, enabling MLLMs to learn from their successful and failed experiences. Following a grow-and-refine principle, the system incrementally accumulates and updates multimodal semantic knowledge -- preserving stable, generalizable strategies while avoiding catastrophic forgetting. Across six multimodal benchmarks, ViLoMem consistently improves pass@1 accuracy and substantially reduces repeated visual and logical errors. Ablations confirm the necessity of dual-stream memory with explicit distraction-hallucination separation, demonstrating the value of error-aware multimodal memory for lifelong and cross-domain agentic learning. Our project page is available at https://weihao-bo.github.io/ViLoMeo-page.

cs.AI↗

arXiv2025

DETree: DEtecting Human-AI Collaborative Texts via Tree-Structured Hierarchical Representation Learning

Detecting AI-involved text is essential for combating misinformation, plagiarism, and academic misconduct. However, AI text generation includes diverse collaborative processes (AI-written text edited by humans, human-written text edited by AI, and AI-generated text refined by other AI), where various or even new LLMs could be involved. Texts generated through these varied processes exhibit complex characteristics, presenting significant challenges for detection. Current methods model these processes rather crudely, primarily employing binary classification (purely human vs. AI-involved) or multi-classification (treating human-AI collaboration as a new class). We observe that representations of texts generated through different processes exhibit inherent clustering relationships. Therefore, we propose DETree, a novel approach that models the relationships among different processes as a Hierarchical Affinity Tree structure, and introduces a specialized loss function that aligns text representations with this tree. To facilitate this learning, we developed RealBench, a comprehensive benchmark dataset that automatically incorporates a wide spectrum of hybrid texts produced through various human-AI collaboration processes. Our method improves performance in hybrid text detection tasks and significantly enhances robustness and generalization in out-of-distribution scenarios, particularly in few-shot learning conditions, further demonstrating the promise of training-based approaches in OOD settings. Our code and dataset are available at https://github.com/heyongxin233/DETree.

cs.CL↗

arXiv2025

OpenSN: An Open Source Library for Emulating LEO Satellite Networks

Low-earth-orbit (LEO) satellite constellations (e.g., Starlink) are becoming a necessary component of future Internet. There have been increasing studies on LEO satellite networking. It is a crucial problem how to evaluate these studies in a systematic and reproducible manner. In this paper, we present OpenSN, i.e., an open source library for emulating large-scale satellite network (SN). Different from Mininet-based SN emulators (e.g., LeoEM), OpenSN adopts container-based virtualization, thus allows for running distributed routing software on each node, and can achieve horizontal scalability via flexible multi-machine extension. Compared to other container-based SN emulators (e.g., StarryNet), OpenSN streamlines the interaction with Docker command line interface and significantly reduces unnecessary operations of creating virtual links. These modifications improve emulation efficiency and vertical scalability on a single machine. Furthermore, OpenSN separates user-defined configuration from container network management via a Key-Value Database that records the necessary information for SN emulation. Such a separation architecture enhances the function extensibility. To sum up, OpenSN exhibits advantages in efficiency, scalability, and extensibility, thus is a valuable open source library that empowers research on LEO satellite networking. Experiment results show that OpenSN constructs mega-constellations 5X-10X faster than StarryNet, and updates link state 2X-4X faster than LeoEM. We also verify the scalability of OpenSN by successfully emulating the five-shell Starlink constellation with a total of 4408 satellites.

cs.NI↗

arXiv2025

Math Blind: Failures in Diagram Understanding Undermine Reasoning in MLLMs

Diagrams represent a form of visual language that encodes abstract concepts and relationships through structured symbols and their spatial arrangements. Unlike natural images, they are inherently symbolic, and entirely artificial. They thus pose unique challenges for Multimodal Large Language Models (MLLMs) distinct from natural image processing. Recent studies have shown that MLLMs often exhibit flawed reasoning and hallucinations when handling diagram inputs. We investigate here whether these limitations stem from shortcomings in the models' ability to interpret diagrams themselves. To this end, we develop a diagnostic test suite that isolates perception from reasoning. Our systematic evaluation reveals that MLLMs perform poorly on basic perceptual tasks, e.g., shape classification, object counting, relationship identification, and object grounding, with near-zero accuracy on fine-grained grounding. Further analysis shows that weak diagram perception leads to "blind faith in text", where models rely on textual shortcuts rather than visual understanding (that is, they are Math Blind). We hypothesize that enabling models to capture the inherent structural properties of diagrams, represented as graphs of primitives and their interrelationships, is essential for improving diagram understanding. Experiments with 7B and 32B MLLMs validate this assumption, with models trained on such representations achieving a +79% gain on the grounding task. Crucially, these gains transfer to reasoning, achieving 3-4% cross-suite improvements on three public benchmarks even without additional chain-of-thought reasoning data. Our findings demonstrate that low-level perception supports faithful high-level reasoning in mathematical MLLMs. We provide both methodological frameworks and empirical evidence to guide future research in this direction.

cs.CV↗