arXiv.itsnick.co

an independent field guide to 23 million papers from arXiv & OpenAlex — not affiliated with arXiv.org

Author

Kevin Han

58 papers · 2016–2026

2016: 22017: 02018: 22019: 42020: 32021: 32022: 72023: 42024: 72025: 132026: 13
2016peak 13 in 20252026

Frequent co-authors

Papers

showing 1–25 of 58 RSS

Computer-Vision-Based Vibration Tracking Using a Digital Camera: A Sparse-Optical-Flow-Based Target Tracking Method

Guang-Yu Nie, Saran Srikanth Bodda, Harleen Kaur Sandhu +2 more

Computer Vision and Pattern Recognition · Sensors · · cited 28× · DOI ↗ · more like this ·

Computer-vision-based target tracking is a technology applied to a wide range of research areas, including structural vibration monitoring. However, current target tracking methods suffer from noise in digital image processing. In this paper, a new target tracking method based on the sparse optical flow technique is introduced for improving the accuracy in tracking the target, especially when the target has a large displacement. The proposed method utilizes the Oriented FAST and Rotated BRIEF (ORB) technique which is based on FAST (Features from Accelerated Segment Test), a feature detector, and BRIEF (Binary Robust Independent Elementary Features), a binary descriptor. ORB maintains a variety of keypoints and combines the multi-level strategy with an optical flow algorithm to search the keypoints with a large motion vector for tracking. Then, an outlier removal method based on Hamming…

Building an Integrated Mobile Robotic System for Real-Time Applications in Construction

Khashayar Asadi, Hariharan Ramshankar, Harish Pullagurla +7 more

cs.RO · · cited 23× · arXiv ↗ · more like this ·

One of the major challenges of a real-time autonomous robotic system for construction monitoring is to simultaneously localize, map, and navigate over the lifetime of the robot, with little or no human intervention. Past research on Simultaneous Localization and Mapping (SLAM) and context-awareness are two active research areas in the computer vision and robotics communities. The studies that integrate both in real-time into a single modular framework for construction monitoring still need further investigation. A monocular vision system and real-time scene understanding are computationally heavy and the major state-of-the-art algorithms are tested on high-end desktops and/or servers with a high CPU- and/or GPU- computing capabilities, which affect their mobility and deployment for real-world applications. To address these challenges and achieve automation, this paper proposes an…

An Efficient Real-Time Object Detection Framework on Resource-Constricted Hardware Devices via Software and Hardware Co-design

Mingshuo Liu, Shiyi Luo, Kevin Han +3 more

Computer Vision and Pattern Recognition · · cited 19× · DOI ↗ · more like this ·

The fast development of object detection techniques has attracted attention to developing efficient Deep Neural Networks (DNNs). However, the current state-of-the-art DNN models can not provide a balanced solution among accuracy, speed, and model size. This paper proposes an efficient real-time object detection framework on resource-constricted hardware devices through hardware and software co-design. The Tensor Train (TT) decomposition is proposed for compressing the YOLOv5 model. By unitizing the unique characteristics given by the TT decomposition, we develop an efficient hardware accelerator based on FPGA devices. Experimental results show that the proposed method can significantly reduce the model size and improve the execution time.

Automatic and Real-Time Joint Tracking and Three-Dimensional Scanning for a Construction Welding Robot

Doyun Lee, Guang-Yu Nie, Kevin Han

Computer Vision and Pattern Recognition · Journal of Construction Engineering and Management · · cited 15× · DOI ↗ · more like this ·

Although welding is one of the essential steel fabrication processes, the American Welding Society expects that the labor shortage in the United States will reach a deficit of 360,000 welders by 2027. Developing an automatic robotic welding system could potentially alleviate the labor shortage and provide better welding quality. As a first step, this paper designs a system pipeline that can automatically detect different welding joints and plan and track the joints’ trajectory with the initial point alignment in real time. There are rare studies that could achieve automatic initial point alignment in real time because the laser stripe’s deformation is not obvious at the narrow weld. In this study, the target joint’s endpoints were detected once the joint was detected on live video. Then, the joint trajectory was planned, and the robotic arm automatically aligned with the initial point…

Autonomous Navigation with Collision Avoidance and Position Adjustment for a Robotic Welding System

Doyun Lee, Kevin Han

Computer Vision and Pattern Recognition · · cited 7× · DOI ↗ · more like this ·

The construction industry has been experiencing a severe welder shortage in the US. Therefore, the development of welding robotics that can move around the job site and perform welding is in critical need. Moreover, welding robotics removes health-related issues associated with harmful fumes, elevated temperatures, and strenuous physical labor. It will also reduce the risk of injuries. An autonomous robot operating on a construction site must be able to detect obstacles, navigate to welding material, and then perform welding on the desired material. This paper focuses on developing an integrated unmanned ground vehicle (UGV) system with an automated robotic welding system to provide mobility for the construction site. The autonomous system maps and navigates with collision avoidance using a 3D light detection and ranging (LiDAR) sensor. When the robot has reached the desired position…

Population Interference in Panel Experiments

Kevin Han, Iavor Bojinov, Guillaume Basse

stat.ME · · cited 7× · arXiv ↗ · more like this ·

The phenomenon of population interference, where a treatment assigned to one experimental unit affects another experimental unit's outcome, has received considerable attention in standard randomized experiments. The complications produced by population interference in this setting are now readily recognized, and partial remedies are well known. Much less understood is the impact of population interference in panel experiments where treatment is sequentially randomized in the population, and the outcomes are observed at each time step. This paper proposes a general framework for studying population interference in panel experiments and presents new finite population estimation and inference results. Our findings suggest that, under mild assumptions, the addition of a temporal dimension to an experiment alleviates some of the challenges of population interference for certain estimands. In…

Exploring Chinese and Korean American Teachers’ Perceptions of Their Cultural Identity as Assets and Barriers

Jiayi Wang, KeAysia Aiyanna Lana Jackson, Eui Kyung Kim +1 more

Sociology and Political Science · Behavioral Sciences · · cited 5× · DOI ↗ · more like this ·

The representation of Asian American teachers in schools is a unique asset in promoting diversity and advocating for social justice. However, they also face various barriers that negatively affect their work experiences. The diverse nature of the Asian American population is likely to yield different assets and barriers among teachers from different ethnic groups. To gain insight into how Chinese and Korean American teachers perceive their cultural identity in their teaching practices, we utilized a qualitative approach by conducting semi-structured interviews with nine teachers who identified as Chinese and/or Korean American in K-12 public schools in California. Findings based on thematic analyses of the interview data revealed several assets, including empathy and understanding towards minority groups, more exposure to diversity, increased representation, and shared language skills.…

Analysis of two pedagogical approaches to foster discipline integrations in an educational data mining class using communities of practice

Brayan Díaz, Collin Lynch, César Delgado +1 more

Computer Science Applications · International Journal of STEM Education · · cited 4× · DOI ↗ · more like this ·

This paper describes research into two pedagogical approaches to foster transdisciplinarity in a graduate engineering course that involves education and computer science. Leveraging the Communities of Practice framework, we examine how students majoring in computer science can integrate new knowledge from education and computer science to engage in an educational data mining project. The first course iteration sought to connect students from education and computer science disciplines through a blend of problem-based learning and traditional lectures. The second course iteration involved computer science students only, but included two instructors, one from computer science and the other from education. To evaluate these approaches, we conducted multiple student interviews and classroom observations. We found that pursuing interdisciplinary through student brokers had a localized student…

Autonomous Binarized Focal Loss Enhanced Model Compression Design Using Tensor Train Decomposition

Mingshuo Liu, Shiyi Luo, Kevin Han +2 more

Computational Mathematics · Micromachines · · cited 4× · DOI ↗ · more like this ·

Deep learning methods have exhibited the great capacity to process object detection tasks, offering a practical and viable approach in many applications. When researchers have advanced deep learning models to improve their performance, the model derived from the algorithmic improvement may itself require complementary increases in computational and power demands. Recently, model compression and pruning techniques have received more attention to promote the wide employment of the DNN model. Although these techniques have achieved a remarkable performance, the class imbalance issue during the mode compression process does not vanish. This paper exploits the Autonomous Binarized Focal Loss Enhanced Model Compression (ABFLMC) model to address the issue. Additionally, our proposed ABFLMC can automatically receive the dynamic difficulty term during the training process to improve performance…

Hardware Oriented Strip-wise Optimization (HOSO) Framework for Efficient Deep Neural Network

Xiaotian Ma, Kevin Han, Yucheng Yang +2 more

Computer Vision and Pattern Recognition · · cited 4× · DOI ↗ · more like this ·

Machine learning methods combined with other computer science technologies have exhibited great potential and unique opportunities to enhance computer vision tasks, offering an effective alternative and a viable method to image processing studies. However, current state-of-the-art object detection methods are either accuracy-oriented within a large model or speed-oriented within a lightweight model. Although some techniques are proposed to compress and accelerate neural networks, most algorithms mainly aim at achieving optimal goals on accuracy and model size based on evaluation in software and lead to a lack of consideration of hardware platforms.To tackle this challenge, in this work, we propose an Hard-ware Oriented Strip-wise Optimization (HOSO) framework, an efficient method that can consider target hardware platforms to compress the machine learning model while retaining high…

An Efficient Video Prediction Recurrent Network using Focal Loss and Decomposed Tensor Train for Imbalance Dataset

Mingshuo Liu, Kevin Han, Shiyi Luo +5 more

Computational Mathematics · · cited 3× · DOI ↗ · more like this ·

Nowadays, from companies to academics, researchers across the world are interested in developing recurrent neural networks due to their incredible feats in various applications, such as speech recognition, video detection, predictions, and machine translation. However, the advantages of recurrent neural networks accompanied by high computational and power demands, which are a major design constraint for electronic devices with limited resources used in such network implementations. Optimizing the recurrent neural networks, such as model compression, is crucial to ensure the broad deployment of recurrent neural networks and promote recurrent neural networks for implementing most resource-constrained scenarios. Among many techniques, tensor train (TT) decomposition is considered an up-and-coming technology. Although our previous efforts have achieved 1) expanding limits of many…

A scaffolding model for designing and implementing work-integrated learning experiences based on the analysis of the university and company's arrangements

Brayan Díaz, César Delgado, Kevin Han +1 more

Education · Higher Education · · cited 2× · DOI ↗ · more like this ·

Abstract In response to the lack of field experience perceived by employers and recent STEM graduates, Work-Integrated Learning (WIL) arrangements have become more common. However, not all WIL experiences are valuable. Indeed, having students in companies without establishing protocols to protect their participation, meaningful interaction, and psychological safety could be counterproductive to non-traditional students. Furthermore, WIL experiences must be designed purposefully to ensure equitable learning experiences for all students. This paper applies an inductive thematic analysis of interviews with 19 current and 4 former students, interviews with current and former workers at host worksites, 76 course surveys, and a total of 15 memos collected from class and workplace observations from graduate engineering classes designed under WIL arrangements between 2017 and 2022. Using…

Detecting Interference in A/B Testing with Increasing Allocation

Kevin Han, Shuangning Li, Jialiang Mao +1 more

stat.ME · · cited 2× · arXiv ↗ · more like this ·

In the past decade, the technology industry has adopted online randomized controlled experiments (a.k.a. A/B testing) to guide product development and make business decisions. In practice, A/B tests are often implemented with increasing treatment allocation: the new treatment is gradually released to an increasing number of units through a sequence of randomized experiments. In scenarios such as experimenting in a social network setting or in a bipartite online marketplace, interference among units may exist, which can harm the validity of simple inference procedures. In this work, we introduce a widely applicable procedure to test for interference in A/B testing with increasing allocation. Our procedure can be implemented on top of an existing A/B testing platform with a separate flow and does not require a priori a specific interference mechanism. In particular, we introduce two…

Estimating Treatment Effects Using Observational Data and Experimental Data with Non-overlapping Support

Kevin Han

stat.ME · · cited 2× · arXiv ↗ · more like this ·

When estimating treatment effects, the golden standard is to conduct a randomized experiment and then contrast outcomes associated with the treatment group and the control group. However, in many cases, randomized experiments are either conducted with a much smaller scale compared to the size of the target population or accompanied with certain ethical issues and thus hard to implement. Therefore, researchers usually rely on observational data to study causal connections. The downside is that the unconfoundedness assumption, the key to validate the use of observational data is hard to verify and almost always violated. Hence, any conclusion drawn from observational data should be further analyzed with great care. Given the richness of observational data and usefulness of experimental data, researchers hope to develop credible method to combine the strength of the two. In this paper, we…

Segmenting Objectiveness and Task-awareness Unknown Region for Autonomous Driving

Mi Zheng, Guanglei Yang, Zitong Huang +3 more

cs.CV · · cited 1× · arXiv ↗ · more like this ·

With the emergence of transformer-based architectures and large language models (LLMs), the accuracy of road scene perception has substantially advanced. Nonetheless, current road scene segmentation approaches are predominantly trained on closed-set data, resulting in insufficient detection capabilities for out-of-distribution (OOD) objects. To overcome this limitation, road anomaly detection methods have been proposed. However, existing methods primarily depend on image inpainting and OOD distribution detection techniques, facing two critical issues: (1) inadequate consideration of the objectiveness attributes of anomalous regions, causing incomplete segmentation when anomalous objects share similarities with known classes, and (2) insufficient attention to environmental constraints, leading to the detection of anomalies irrelevant to autonomous driving tasks. In this paper, we propose…

Gaussian and Non-Gaussian Universality of Data Augmentation

Kevin Han Huang, Peter Orbanz, Morgane Austern

cs.LG · · cited 1× · arXiv ↗ · more like this ·

We provide universality results that quantify how data augmentation affects the variance and limiting distribution of estimates through simple surrogates, and analyze several specific models in detail. The results confirm some observations made in machine learning practice, but also lead to unexpected findings: Data augmentation may increase rather than decrease the uncertainty of estimates, such as the empirical prediction risk. It can act as a regularizer, but fails to do so in certain high-dimensional problems, and it may shift the double-descent peak of an empirical risk. Overall, the analysis shows that several properties data augmentation has been attributed with are not either true or false, but rather depend on a combination of factors -- notably the data distribution, the properties of the estimator, and the interplay of sample size, number of augmentations, and dimension. As…

Generator control action classification based on localized voltage measurements

Shamina Hossain‐McKenzie, Kevin Han, Katherine Davis

Signal Processing · · cited 1× · DOI ↗ · more like this ·

Voltage disturbances in power systems are caused by various events such as faults or control actions. It is important to accurately identify them for fast remedial response and damage prevention. Recently, data mining techniques have been applied in power system research, usually from the perspective of the entire system. The analyzed data include measurements from transmission buses, substations, and other major components in the network. This big data analysis is useful and enables powerful insight but studies on small, local areas of the larger system needs to be explored further. By focusing on single components such as generators, we can provide concrete and detailed insights to individual operators. For instance, direct impact on the local area can be studied after an event - allowing for better understanding of its propagation effects throughout the entire system. This provides…

Beyond Semantic IDs: Atomic Creator Tokens for Large-Scale Generative Recommendation

Zhuosheng Liu, Yihan Wu, Ziyun Xu +8 more

cs.AI · · cited 0× · arXiv ↗ · more like this ·

Recommending creators that a user does not yet follow is a cold-start problem: from hundreds of thousands of popular creators and only a short trail of prior engagements, the system must predict the next creator worth surfacing. The prevailing generative recommendation recipe tokenizes each creator into a \emph{Semantic ID} (SID) by quantizing a frozen embedding and trains a language model to generate SIDs. We show that this SID layer, not the language model, is the binding constraint. The creator-to-SID map is many-to-one, so colliding creators must be resolved by a popularity tie-break that caps creator-level recall and favors already-popular creators, and the frozen quantizer stops the training gradient from reaching the creator representation. We propose Atomic Creator Token (ACT), which gives every creator its own vocabulary token trained end-to-end. The map is bijective, so…

Baseline and Incremental: User, Item, and Request Debiasing in Instagram Feed Ranking

Hsin-Chiang Chen, Hao Wang, Gedi Zhou +23 more

cs.IR · · cited 0× · arXiv ↗ · more like this ·

A feed ranker scores candidates with a value model, a weighted sum of engagement predictions. Each prediction conflates a baseline, what this viewer, item or request would produce anyway, with the incremental amount by which this candidate exceeds it. The literature mostly removes the baseline. We argue that a multi-signal value model should keep the raw prediction and add the incremental part as a separately weighted term. On Instagram Feed, three debiasing families deployed independently, for viewer habit, item popularity and request context, are one parameterised operator, and a design procedure picks a point in it. Media-type (format) bias goes instead to a debiased label chosen by the same procedure. In a controlled four-arm ablation of deployed Feed terms (neither term, each alone, both), the two-term arm beats both single-term arms at production weights on point estimates in 11…

TreeFlex: Agents Navigating Their Own Search Trees

Dongzhuo Li, Fangzhou Cheng, Weiwei Li +8 more

cs.AI · · cited 0× · arXiv ↗ · more like this ·

A ReAct-style agent accumulates observations, intermediate results, and failed attempts in a single context. As this context grows, stale and off-path information consumes its limited capacity and obscures relevant evidence, making focused reasoning increasingly difficult. We introduce TreeFlex, a framework in which the agent's own policy decides where to search. The agent records each intermediate state, with a short summary, as a node in a search tree maintained by the harness. The policy is a single language model that acts in two roles. In the Reflection role, it has a global view of the tree through these summaries and picks a node to resume, together with an instruction. In the Execution role, the harness restores the path that led to that node, and the model carries out the instruction with the usual task tools until it saves a new node or proposes a solution, which Reflection…

Persistent Negatives for Adversarial Black-Box On-Policy Distillation

Haixu Ma, Saad Lahrichi, Weiwei Li +11 more

cs.CL · · cited 0× · arXiv ↗ · more like this ·

Black-box On-Policy Distillation (OPD) seeks to improve a student from its own generations when the teacher provides sampled responses but not token probabilities. Adversarial distillation offers one route: it learns a discriminator over prompt-matched teacher and student responses and uses its score as the policy reward. However, sampling discriminator negatives from the latest student at each step couples the learned reward to a negative distribution that changes after every policy update. We address this moving-target problem with persistent-negative adversarial distillation, a live-pool method that replaces a fraction of each discriminator batch with historical, prompt-matched teacher--student comparisons. Under matched discriminator compute, historical comparisons train the discriminator, while GRPO remains on-policy with fresh student responses. Our analysis identifies the…

Enhancing Construction Requirements Documents Analytics through Integrating Ontology and Large Language Model

Liannian Wang, Jeongbin Hwang, Kevin K. Han +1 more

Artificial Intelligence · Journal of Management in Engineering · · cited 0× · DOI ↗ · more like this ·

Abstract Ineffective management of construction requirements results in delays, cost overruns, and compliance challenges due to extensive documentation and manual processes. While digital engineering improves automation, its effectiveness depends on accurate information extraction. Existing large language model (LLM)-based approaches have improved construction text analytics, and construction ontologies have supported structured knowledge representation. However, ontology-guided LLM workflows for practical sentence-level extraction and structuring of construction requirements remain underexplored. This study addresses these challenges by integrating a new ontology with an LLM, using the ontology for structured knowledge representation and the LLM for advanced language processing. The proposed ontology systematically captures the foundational aspects of construction requirements across…

DRT: Dense Reasoning Trace for Efficient and Grounded Multimodal Reasoning

Wan Xu, Yuanfan Guo, Kevin Han +2 more

cs.CV · · cited 0× · arXiv ↗ · more like this ·

Despite the remarkable progress in Multimodal Large Language Models (MLLMs), prevailing Chain-of-Thought (CoT) paradigms remain confined to the natural-language expression space. Consequently, they inherently incur excessive linguistic overhead, leading to information dilution and weak visual grounding. To address this challenge, we propose Dense Reasoning Trace (DRT), a paradigm that departs from natural-language-centered CoT by expressing reasoning as compact structured traces, which include concise intermediate states with symbolic connectors and disentangle visual observations from logical deductions. First, we introduce the Dense Trace Initialization to internalize the DRT reasoning mode into the model, substantially improving token efficiency while preserving visual evidence. To further enable the model to faithfully capture the logical relations within traces, we propose the…

Head-mounted first-person view camera-based worker action recognition using visual SLAM

Jeongbin Hwang, Ahmed Afifi, Kevin Han

Computer Vision and Pattern Recognition · Automation in Construction · · cited 0× · DOI ↗ · more like this ·

Since construction workforce management is crucial for productivity and safety, recognizing workers’ actions on site has become important. Researchers have explored various sensors, including cameras, for worker action recognition. To directly capture egocentric body and head movements from the worker’s perspective, this paper proposes an action recognition approach using a head-mounted first-person view (FPV) camera. The proposed approach defines and analyzes seven body movement types and five head movement types, supported by visual simultaneous localization and mapping (SLAM)-based FPV camera pose estimation and 3D head pose–based action recognition. It achieved accuracy of 94.52% (body: 97.08%, head: 92.83%) indoors and 89.31% (body: 90.62%, head: 87.99%) outdoors, with macro F1-scores of 87.58% (body: 95.04%, head: 80.13%) and 78.14% (body: 82.53%, head: 73.75%), respectively.…

A Qualitative Examination of School Climate among Asian American K-12 Teachers in California

Eui Kyung Kim, Kevin Han, Jiayi Wang +1 more

Education · School Psychology Review · · cited 0× · DOI ↗ · more like this ·

Teachers are vital members of the school community who play a crucial role in creating a positive school environment for everyone, including students, families, and themselves. However, despite Asian American students representing 6% of the K-12 student population in the US, only 2% of teachers are reportedly Asian American, indicating a notable underrepresentation in the teaching profession. The existing literature on teacher school climate often does not address teachers’ racial and ethnic identity, despite its importance on their experiences and perceptions of the school environment. To identify factors that indicate a positive perception of school climate for Asian American teachers, this study conducted semi-structured interviews with 12 Asian American teachers in California K-12 schools. Thematic analyses revealed five school climate factors: teacher relations, work conditions…