arXiv subjects

Yu-Hsuan Lin

Publications and source records attributed to Yu-Hsuan Lin.

6 recordsLinked to original sources

arXiv2025

Ordinal Scale Traffic Congestion Classification with Multi-Modal Vision-Language and Motion Analysis

Accurate traffic congestion classification is essential for intelligent transportation systems and real-time urban traffic management. This paper presents a multimodal framework combining open-vocabulary visual-language reasoning (CLIP), object detection (YOLO-World), and motion analysis via MOG2-based background subtraction. The system predicts congestion levels on an ordinal scale from 1 (free flow) to 5 (severe congestion), enabling semantically aligned and temporally consistent classification. To enhance interpretability, we incorporate motion-based confidence weighting and generate annotated visual outputs. Experimental results show the model achieves 76.7 percent accuracy, an F1 score of 0.752, and a Quadratic Weighted Kappa (QWK) of 0.684, significantly outperforming unimodal baselines. These results demonstrate the framework's effectiveness in preserving ordinal structure and leveraging visual-language and motion modalities. Future enhancements include incorporating vehicle sizing and refined density metrics.

cs.CV↗

arXiv2025

LLM Inference Enhanced by External Knowledge: A Survey

Recent advancements in large language models (LLMs) have enhanced natural-language reasoning. However, their limited parametric memory and susceptibility to hallucination present persistent challenges for tasks requiring accurate, context-based inference. To overcome these limitations, an increasing number of studies have proposed leveraging external knowledge to enhance LLMs. This study offers a systematic exploration of strategies for using external knowledge to enhance LLMs, beginning with a taxonomy that categorizes external knowledge into unstructured and structured data. We then focus on structured knowledge, presenting distinct taxonomies for tables and knowledge graphs (KGs), detailing their integration paradigms with LLMs, and reviewing representative methods. Our comparative analysis further highlights the trade-offs among interpretability, scalability, and performance, providing insights for developing trustworthy and generalizable knowledge-enhanced LLMs.

cs.CL↗

arXiv2024

SMLT-MUGC: Small, Medium, and Large Texts -- Machine versus User-Generated Content Detection and Comparison

Large language models (LLMs) have gained significant attention due to their ability to mimic human language. Identifying texts generated by LLMs is crucial for understanding their capabilities and mitigating potential consequences. This paper analyzes datasets of varying text lengths: small, medium, and large. We compare the performance of machine learning algorithms on four datasets: (1) small (tweets from Election, FIFA, and Game of Thrones), (2) medium (Wikipedia introductions and PubMed abstracts), and (3) large (OpenAI web text dataset). Our results indicate that LLMs with very large parameters (such as the XL-1542 variant of GPT2 with 1542 million parameters) were harder (74%) to detect using traditional machine learning methods. However, detecting texts of varying lengths from LLMs with smaller parameters (762 million or less) can be done with high accuracy (96% and above). We examine the characteristics of human and machine-generated texts across multiple dimensions, including linguistics, personality, sentiment, bias, and morality. Our findings indicate that machine-generated texts generally have higher readability and closely mimic human moral judgments but differ in personality traits. SVM and Voting Classifier (VC) models consistently achieve high performance across most datasets, while Decision Tree (DT) models show the lowest performance. Model performance drops when dealing with rephrased texts, particularly shorter texts like tweets. This study underscores the challenges and importance of detecting LLM-generated texts and suggests directions for future research to improve detection methods and understand the nuanced capabilities of LLMs.

cs.CL↗

arXiv2020

Unambiguous Electrical Detection of Spin-Charge Conversion in Lateral Spin-Valves

Efficient detection of spin-charge conversion is crucial for advancing our understanding of emergent phenomena in spin-orbit-coupled nanostructures. Here, we provide proof of principle of an electrical detection scheme of spin-charge conversion that enables full disentanglement of competing spin-orbit coupling transport phenomena in diffusive lateral channels i.e. the inverse spin Hall effect (ISHE) and the spin galvanic effect (SGE). A suitable detection geometry in an applied oblique magnetic field is shown to provide direct access to spin-charge transport coefficients through a simple symmetry analysis of the output non-local resistance. The scheme is robust against tilting of the spin-injector magnetization, disorder and spurious non-spin related contributions to the non-local signal, and can be used to probe spin-charge conversion effects in both spin-valve and hybrid optospintronic devices.

cond-mat.mes-hall↗

arXiv2019

Zero-field spin-orbit-torque switching driven by magnetic spin Hall effect

Spin Hall effect plays an essential role in generating spin current from the injected charge current, following the Dyakonov-Perel rule that the directions of charge flow, spin flow and spin polarization are mutually perpendicular to each other. Recently, its generalization from an antiferromagnet, so-called magnetic spin Hall effect, has been studied and verified by measuring anomalous spin accumulations. Here, we investigate the magnetic spin Hall effect in bilayer materials made of a heavy metal and an antiferromagnet. The spin current generated by the magnetic spin Hall effect accomplishes spin-orbit-torque switching for ferromagnetic magnetization and exchange bias concurrently without any external magnetic field. The switching mechanism crucially relies on the non-collinear spin texture in the antiferromagnet, capable of generating symmetry-breaking components in the spin-current tensor so that the external magnetic field is no longer necessary. The zero-field concurrent switching of magnetization and exchange bias is a significant technological breakthrough. Furthermore, our findings pave the way to explore the magnetic spin Hall effects in various spin textures through spin-orbit-torque switching.

cond-mat.mtrl-sci↗

arXiv2019

Theory of Spin Injection in Two-dimensional Metals with Proximity-Induced Spin-Orbit Coupling

Spin injection is a powerful experimental probe into a wealth of nonequilibrium spin-dependent phenomena displayed by materials with spin-orbit coupling (SOC). Here, we develop a theory of coupled spin-charge diffusive transport in two-dimensional spin-valve devices. The theory describes a realistic proximity-induced SOC with both spatially uniform and random components of the SOC due to adatoms and imperfections, and applies to the two dimensional electron gases found in two-dimensional materials and van der Walls heterostructures. The various charge-to-spin conversion mechanisms known to be present in diffusive metals, including the spin Hall effect and several mechanisms contributing current-induced spin polarization are accounted for. Our analysis shows that the dominant conversion mechanisms can be discerned by analyzing the nonlocal resistance of the spin-valve for different polarizations of the injected spins and as a function of the applied in-plane magnetic field.

cond-mat.mes-hall↗