arXiv subjects

Yiwei Sun

Publications and source records attributed to Yiwei Sun.

At least 19 recordsLinked to original sources

arXiv2026

Two-Phase Simulated Annealing for Equitable Team Formation: Eliminating Complaints in Large Engineering Cohorts

Contribution: This paper presents a novel two-phase algorithmic approach that decouples preference satisfaction from fairness optimization in student team formation, achieving both objectives without compromise. The method applies simulated annealing -- a core materials science technique -- to an educational challenge, demonstrating pedagogical integration of administrative processes. Background: Forming effective teams in large engineering cohorts (100+ students) requires balancing student preferences, academic fairness, and demographic diversity. Existing tools either optimize for fairness while ignoring preferences (CATME, Team-Anneal) or accommodate preferences while compromising balance (self-selection), leaving complaint rates at 5--35%. Intended Outcomes: Eliminate formal complaints, achieve near-zero GPA variance between teams, prevent gender isolation, and maintain high preference satisfaction while creating a scalable, reproducible solution applicable across engineering programs. Application Design: Phase 1 forms fixed triads through graph-theoretic clustering that maximizes mutual preferences, preserving social bonds. Phase 2 employs simulated annealing to pair triads into teams of six while optimizing GPA variance, gender balance, and size constraints. This decomposition mirrors hierarchical optimization in materials processing. Findings: Deployed across 238 students, the algorithm eliminated formal complaints entirely (vs >30% baseline), achieved GPA variance of 0.005 (vs. historical mean 9.74), eliminated gender-isolated individuals, and maintained 94.3% preference satisfaction. Validation against 82 historical grouping instances (1,538 teams, 6 academic years) confirmed significant improvement over conventional methods.

cs.CY↗

arXiv2026

Detective scaffolding for within-session reasoning development: a three-phase framework evaluated in polymer engineering and pre-university outreach

This paper presents a detective scaffolding framework -- a three-phase instructional sequence (Hypothesis Activation -> Evidence Structuring -> Causal Integration) in which engineering students investigate a realistic industrial defect scenario using staged in-class polls as designed evidence probes. Unlike conventional uses of student response systems for engagement, the framework positions each poll as an Evidence-Centred Design instrument targeting a specific reasoning capability. In the primary implementation, 80 Year~3 polymer engineering students progressed from prior-knowledge-driven misconception (71% attributing defects to temperature) to complete root-cause convergence (100\% identifying humidity; Fisher's exact test, $p < .001$) across four sequenced prompts within a single 90-minute lecture slot. A dual-accuracy analysis revealed that at one intermediate stage, textbook-correct and analytically valid responses diverged, illustrating why conventional scoring can misrepresent reasoning quality. In a transferability study, 26 Year~12 students with no engineering background achieved identical root-cause identification rates across two adapted scenarios, with significant gains in data-analysis confidence and AI explanation ability. The results suggest that the pedagogical structure, rather than disciplinary content, drives the convergence effect, implying portability across disciplines and educational levels.

physics.ed-ph↗

arXiv2025

Decoupling Perception from Reasoning for Hallucination-Resistant Video Understanding

Video Large Language Models improve reasoning over complex videos by generating intermediate reasoning text. However, reliable reasoning depends on accurate video perception. In existing approaches, perception evidence is intertwined with reasoning text, making it difficult to directly supervise the perception process. We argue that reliable supervision requires explicitly separating perception evidence from reasoning so that perception can be verified independently. To supervise perception directly, we propose Decoupled Perception and Logic (DPL), which represents perception as fixed-format evidence units containing timestamps and visual descriptions. This structured representation enables direct extraction of perception content and simplifies alignment between video segments and reward evaluation. Building on DPL, we introduce a perception reward that encourages both hallucination resistance and perception-based reasoning. An Factual-Aware Evaluator (FAE) provides anti-hallucination scores and achieves hallucination evaluation performance comparable to GPT-4o. In addition, we validate reasoning consistency by feeding perception results and questions into a reference model. Experiments show that, by providing reliable process rewards, Video-DPL consistently improves post-training performance at both 3B and 7B scales, while delivering higher data efficiency.

cs.CV↗

arXiv2025

RegionRAG: Region-level Retrieval-Augmented Generation for Visual Document Understanding

Multi-modal Retrieval-Augmented Generation (RAG) has become a critical method for empowering LLMs by leveraging candidate visual documents. However, current methods consider the entire document as the basic retrieval unit, introducing substantial irrelevant visual content in two ways: 1) Relevant documents often contain large regions unrelated to the query, diluting the focus on salient information; 2) Retrieving multiple documents to increase recall further introduces redundant and irrelevant documents. These redundant contexts distract the model's attention and further degrade the performance. To address this challenge, we propose RegionRAG, a novel framework that shifts the retrieval paradigm from the document level to the region level. During training, we design a hybrid supervision strategy from both labeled data and unlabeled data to pinpoint relevant patches. During inference, we propose a dynamic pipeline that intelligently groups salient patches into complete semantic regions. By delegating the task of identifying relevant regions to the retriever, RegionRAG enables the generator to focus solely on concise, query-relevant visual content, improving both efficiency and accuracy. Experiments on six benchmarks demonstrate that RegionRAG achieves state-of-the-art performance. It improves retrieval accuracy by 10.02% in R@1 on average, and boosts question answering accuracy by 3.56% while using only 71.42% visual tokens compared with prior methods.

cs.CV↗

arXiv2025

Hierarchical Bayesian Knowledge Tracing in Undergraduate Engineering Education

Educators teaching entry-level university engineering modules face the challenge of identifying which topics students find most difficult and how to support diverse student needs effectively. This study demonstrates a rigorous yet interpretable statistical approach -- hierarchical Bayesian modeling -- that leverages detailed student response data to quantify both skill difficulty and individual student abilities. Using a large-scale dataset from an undergraduate Statics course, we identified clear patterns of skill mastery and uncovered distinct student subgroups based on their learning trajectories. Our analysis reveals that certain concepts consistently present challenges, requiring targeted instructional support, while others are readily mastered and may benefit from enrichment activities. Importantly, the hierarchical Bayesian method provides educators with intuitive, reliable metrics without sacrificing predictive accuracy. This approach allows for data-informed decisions, enabling personalized teaching strategies to improve student engagement and success. By combining robust statistical methods with clear interpretability, this study equips educators with actionable insights to better support diverse learner populations.

cs.CY↗

arXiv2025

From Evaluation to Defense: Advancing Safety in Video Large Language Models

While the safety risks of image-based large language models (Image LLMs) have been extensively studied, their video-based counterparts (Video LLMs) remain critically under-examined. To systematically study this problem, we introduce VideoSafetyEval - a large-scale, real-world benchmark for Video LLM safety, which comprises 11.4k video-query pairs and spans 19 principal risk categories. Based on this, we reveal that integrating video modality degrades safety performance by an average of 34.2%, thereby exposing systemic risks in multimodal attack exploitation. To address this vulnerability, we propose VideoSafety-R1, a dual-stage framework achieving unprecedented safety gains through three innovations: (1) the VideoSafetyThinking dataset contains 46k video-query-thinking response triplets; (2) Alarm Token-Guided Safety Fine-Tuning (AT-SFT) injects learnable alarm tokens into visual and textual sequences, enabling explicit harm perception across modalities via multitask objectives; and (3) safety-guided GRPO enhances defensive reasoning through dynamic policy optimization with rule-based rewards derived from dual-modality verification. These components synergize to shift safety alignment from harm perception to active reasoning. The framework achieves a 71.1% improvement on VSE-HH, and improves by 59.1%, 44.3%, and 15.0% on the image safety datasets MMBench, VLGuard, and FigStep, respectively. Our code and dataset are available at https://github.com/Emiya-syw/VideoSafety-R1.git. Note: This paper contains harmful language and image examples, and reader discretion is recommended.

cs.CV↗

arXiv2024

Externally Valid Selection of Experimental Sites via the k-Median Problem

We present a decision-theoretic justification for viewing the question of how to best choose where to experiment in order to optimize external validity as a $k$-median problem, a popular problem in computer science and operations research. We present conditions under which minimizing the worst-case, welfare-based regret among all nonrandom schemes that select $k$ sites to experiment is approximately equal - and sometimes exactly equal - to finding the k most central vectors of baseline site-level covariates. The k-median problem can be formulated as a linear integer program. Two empirical applications illustrate the theoretical and computational benefits of the suggested procedure.

econ.EM↗

arXiv2024

Hallucination Mitigation Prompts Long-term Video Understanding

Recently, multimodal large language models have made significant advancements in video understanding tasks. However, their ability to understand unprocessed long videos is very limited, primarily due to the difficulty in supporting the enormous memory overhead. Although existing methods achieve a balance between memory and information by aggregating frames, they inevitably introduce the severe hallucination issue. To address this issue, this paper constructs a comprehensive hallucination mitigation pipeline based on existing MLLMs. Specifically, we use the CLIP Score to guide the frame sampling process with questions, selecting key frames relevant to the question. Then, We inject question information into the queries of the image Q-former to obtain more important visual features. Finally, during the answer generation stage, we utilize chain-of-thought and in-context learning techniques to explicitly control the generation of answers. It is worth mentioning that for the breakpoint mode, we found that image understanding models achieved better results than video understanding models. Therefore, we aggregated the answers from both types of models using a comparison mechanism. Ultimately, We achieved 84.2\% and 62.9\% for the global and breakpoint modes respectively on the MovieChat dataset, surpassing the official baseline model by 29.1\% and 24.1\%. Moreover the proposed method won the third place in the CVPR LOVEU 2024 Long-Term Video Question Answering Challenge. The code is avaiable at https://github.com/lntzm/CVPR24Track-LongVideo

cs.CV↗

arXiv2024

Design principles of nonlinear optical materials for Terahertz lasers

We have investigated both inter-band and intra-band second order nonlinear optical conductivity based on the velocity correlation formalism and the spectral expansion technique. We propose a scenario in which the second order intra-band process is nonzero while the inter-band process is zero. This occurs for a band structure with momentum asymmetry in the Brillouin zone. Very low-energy photons are blocked by the Pauli exclusion principle from participating in the inter-band process; however, they are permitted to participate in the intra-band process, with the band smeared by some impurity scattering. We establish a connection between the inter-band nonlinear optical conductivity in the velocity gauge and the shift vector in the length gauge for a two-band model. Using a quasiclassical kinetic approach, we demonstrate the importance of intra-band transitions in high harmonic generations for the single tilted Dirac cone model and hexagonal warping model. We confirm that the Kramers-Kronig relations break down for the limit case of ($\omega$, $-\omega$) in the nonlinear optical conductivity. Finally, we calculate the superconducting transition temperature of NbN and the dielectric function of AlN, and the resistance of the NbN/AlN junction. The natural non-linearity of the Josephson junction brings a Josephson plasma with frequency in the Terahertz region.

cond-mat.mtrl-sci↗

arXiv2023

Extrapolating Away from the Cutoff in Regression Discontinuity Designs

Canonical RD designs yield credible local estimates of the treatment effect at the cutoff under mild continuity assumptions, but they fail to identify treatment effects away from the cutoff without additional assumptions. The fundamental challenge of identifying treatment effects away from the cutoff is that the counterfactual outcome under the alternative treatment status is never observed. This paper aims to provide a methodological blueprint to identify treatment effects away from the cutoff in various empirical settings by offering a non-exhaustive list of assumptions on the counterfactual outcome. Instead of assuming the exact evolution of the counterfactual outcome, this paper bounds its variation using the data and sensitivity parameters. The proposed assumptions are weaker than those introduced previously in the literature, resulting in partially identified treatment effects that are less susceptible to assumption violations. This approach accommodates both single cutoff and multi-cutoff designs. The specific choice of the extrapolation assumption depends on the institutional background of each empirical application. Additionally, researchers are recommended to conduct sensitivity analysis on the chosen parameter and assess resulting shifts in conclusions. The paper compares the proposed identification results with results using previous methods via an empirical application and simulated data. It demonstrates that set identification yields a more credible conclusion about the sign of the treatment effect.

econ.EM↗

arXiv2023

Effect of Composition on Microstructural Evolution during Homogenization of 7XXX Alloys

The effect of composition on microstructure both at the length scale of the secondary dendrite arm spacing and nano-sized dispersoids during homogenization of Al-Zn-Cu-Mg-Zr alloys has been studied. A comprehensive model that can predict the microstructure at both the length scales has been used for the study. The microstructure predicted has been compared to that for two homogenized samples from a directionally solidified AA7050 sample and a reasonable match has been found. The initial as-cast microstructure for different compositions is calculated using Scheil type solidification from Thermo-Calc. The initial microstructure has a considerable influence on microstructural evolution during homogenization. To take advantage of the decreased solid solubility of Zr in {\alpha}-fcc, cooling rates during solidification must be high enough to prevent precipitation of primary Al3Zr. Under solidification with industrial cooling conditions solute rich alloys leads to fewer dispersoids. Based on the study, an improved composition range of 6-8%Zn, 1-2%Cu, 1-2%Mg and 0.1-0.15%Zr for 7XXX alloys has been proposed.

cond-mat.mtrl-sci↗

arXiv2021

Labeled Data Generation with Inexact Supervision

The recent advanced deep learning techniques have shown the promising results in various domains such as computer vision and natural language processing. The success of deep neural networks in supervised learning heavily relies on a large amount of labeled data. However, obtaining labeled data with target labels is often challenging due to various reasons such as cost of labeling and privacy issues, which challenges existing deep models. In spite of that, it is relatively easy to obtain data with \textit{inexact supervision}, i.e., having labels/tags related to the target task. For example, social media platforms are overwhelmed with billions of posts and images with self-customized tags, which are not the exact labels for target classification tasks but are usually related to the target labels. It is promising to leverage these tags (inexact supervision) and their relations with target classes to generate labeled data to facilitate the downstream classification tasks. However, the work on this is rather limited. Therefore, we study a novel problem of labeled data generation with inexact supervision. We propose a novel generative framework named as ADDES which can synthesize high-quality labeled data for target classification tasks by learning from data with inexact supervision and the relations between inexact supervision and target classes. Experimental results on image and text datasets demonstrate the effectiveness of the proposed ADDES for generating realistic labeled data from inexact supervision to facilitate the target classification task.

cs.LG↗

arXiv2020

Explainable Multivariate Time Series Classification: A Deep Neural Network Which Learns To Attend To Important Variables As Well As Informative Time Intervals

Time series data is prevalent in a wide variety of real-world applications and it calls for trustworthy and explainable models for people to understand and fully trust decisions made by AI solutions. We consider the problem of building explainable classifiers from multi-variate time series data. A key criterion to understand such predictive models involves elucidating and quantifying the contribution of time varying input variables to the classification. Hence, we introduce a novel, modular, convolution-based feature extraction and attention mechanism that simultaneously identifies the variables as well as time intervals which determine the classifier output. We present results of extensive experiments with several benchmark data sets that show that the proposed method outperforms the state-of-the-art baseline methods on multi-variate time series classification task. The results of our case studies demonstrate that the variables and time intervals identified by the proposed method make sense relative to available domain knowledge.

cs.LG↗

arXiv2020

Investigating and Mitigating Degree-Related Biases in Graph Convolutional Networks

Graph Convolutional Networks (GCNs) show promising results for semi-supervised learning tasks on graphs, thus become favorable comparing with other approaches. Despite the remarkable success of GCNs, it is difficult to train GCNs with insufficient supervision. When labeled data are limited, the performance of GCNs becomes unsatisfying for low-degree nodes. While some prior work analyze successes and failures of GCNs on the entire model level, profiling GCNs on individual node level is still underexplored. In this paper, we analyze GCNs in regard to the node degree distribution. From empirical observation to theoretical proof, we confirm that GCNs are biased towards nodes with larger degrees with higher accuracy on them, even if high-degree nodes are underrepresented in most graphs. We further develop a novel Self-Supervised-Learning Degree-Specific GCN (SL-DSGC) that mitigate the degree-related biases of GCNs from model and data aspects. Firstly, we propose a degree-specific GCN layer that captures both discrepancies and similarities of nodes with different degrees, which reduces the inner model-aspect biases of GCNs caused by sharing the same parameters with all nodes. Secondly, we design a self-supervised-learning algorithm that creates pseudo labels with uncertainty scores on unlabeled nodes with a Bayesian neural network. Pseudo labels increase the chance of connecting to labeled neighbors for low-degree nodes, thus reducing the biases of GCNs from the data perspective. Uncertainty scores are further exploited to weight pseudo labels dynamically in the stochastic gradient descent for SL-DSGC. Experiments on three benchmark datasets show SL-DSGC not only outperforms state-of-the-art self-training/self-supervised-learning GCN methods, but also improves GCN accuracy dramatically for low-degree nodes.

cs.LG↗

arXiv2020

Principal Component Analysis Based Broadband Hybrid Precoding for Millimeter-Wave Massive MIMO Systems

Hybrid analog-digital precoding is challenging for broadband millimeter-wave (mmWave) massive MIMO systems, since the analog precoder is frequency-flat but the mmWave channels are frequency-selective. In this paper, we propose a principal component analysis (PCA)-based broadband hybrid precoder/combiner design, where both the fully-connected array and partially-connected subarray (including the fixed and adaptive subarrays) are investigated. Specifically, we first design the hybrid precoder/combiner for fully-connected array and fixed subarray based on PCA, whereby a low-dimensional frequency-flat precoder/combiner is acquired based on the optimal high-dimensional frequency-selective precoder/combiner. Meanwhile, the near-optimality of our proposed PCA approach is theoretically proven. Moreover, for the adaptive subarray, a low-complexity shared agglomerative hierarchical clustering algorithm is proposed to group the antennas for the further improvement of spectral efficiency (SE) performance. Besides, we theoretically prove that the proposed antenna grouping algorithm is only determined by the slow time-varying channel parameters in the large antenna limit. Simulation results demonstrate the superiority of the proposed solution over state-of-the-art schemes in SE, energy efficiency (EE), bit-error-rate performance, and the robustness to time-varying channels. Our work reveals that the EE advantage of adaptive subarray over fully-connected array is obvious for both active and passive antennas, but the EE advantage of fixed subarray only holds for passive antennas.

cs.IT↗

arXiv2020

Ginger Cannot Cure Cancer: Battling Fake Health News with a Comprehensive Data Repository

Nowadays, Internet is a primary source of attaining health information. Massive fake health news which is spreading over the Internet, has become a severe threat to public health. Numerous studies and research works have been done in fake news detection domain, however, few of them are designed to cope with the challenges in health news. For instance, the development of explainable is required for fake health news detection. To mitigate these problems, we construct a comprehensive repository, FakeHealth, which includes news contents with rich features, news reviews with detailed explanations, social engagements and a user-user social network. Moreover, exploratory analyses are conducted to understand the characteristics of the datasets, analyze useful patterns and validate the quality of the datasets for health fake news detection. We also discuss the novel and potential future research directions for the health fake news detection.

cs.SI↗

arXiv2020

Transferring Robustness for Graph Neural Network Against Poisoning Attacks

Graph neural networks (GNNs) are widely used in many applications. However, their robustness against adversarial attacks is criticized. Prior studies show that using unnoticeable modifications on graph topology or nodal features can significantly reduce the performances of GNNs. It is very challenging to design robust graph neural networks against poisoning attack and several efforts have been taken. Existing work aims at reducing the negative impact from adversarial edges only with the poisoned graph, which is sub-optimal since they fail to discriminate adversarial edges from normal ones. On the other hand, clean graphs from similar domains as the target poisoned graph are usually available in the real world. By perturbing these clean graphs, we create supervised knowledge to train the ability to detect adversarial edges so that the robustness of GNNs is elevated. However, such potential for clean graphs is neglected by existing work. To this end, we investigate a novel problem of improving the robustness of GNNs against poisoning attacks by exploring clean graphs. Specifically, we propose PA-GNN, which relies on a penalized aggregation mechanism that directly restrict the negative impact of adversarial edges by assigning them lower attention coefficients. To optimize PA-GNN for a poisoned graph, we design a meta-optimization algorithm that trains PA-GNN to penalize perturbations using clean graphs and their adversarial counterparts, and transfers such ability to improve the robustness of PA-GNN on the poisoned graph. Experimental results on four real-world datasets demonstrate the robustness of PA-GNN against poisoning attacks on graphs. Code and data are available here: https://github.com/tangxianfeng/PA-GNN.

cs.LG↗

arXiv2020

An Ontology-Aware Framework for Audio Event Classification

Recent advancements in audio event classification often ignore the structure and relation between the label classes available as prior information. This structure can be defined by ontology and augmented in the classifier as a form of domain knowledge. To capture such dependencies between the labels, we propose an ontology-aware neural network containing two components: feed-forward ontology layers and graph convolutional networks (GCN). The feed-forward ontology layers capture the intra-dependencies of labels between different levels of ontology. On the other hand, GCN mainly models inter-dependency structure of labels within an ontology level. The framework is evaluated on two benchmark datasets for single-label and multi-label audio event classification tasks. The results demonstrate the proposed solutions efficacy to capture and explore the ontology relations and improve the classification performance.

cs.AI↗