Author

Heng Lin

184 papers · 1986–2026

1986peak 32 in 20252026

Frequent co-authors

Papers

showing 1–25 of 184 RSS

FreeControl: Training-Free Spatial Control of Any Text-to-Image Diffusion Model with Any Condition

Sicheng Mo, Fangzhou Mu, Kuan Heng Lin +4 more

cs.CV · 12 Dec 2023 · cited 131× · arXiv ↗ · more like this ·

Recent approaches such as ControlNet offer users fine-grained spatial control over text-to-image (T2I) diffusion models. However, auxiliary modules have to be trained for each type of spatial condition, model architecture, and checkpoint, putting them at odds with the diverse intents and preferences a human designer would like to convey to the AI models during the content creation process. In this work, we present FreeControl, a training-free approach for controllable T2I generation that supports multiple conditions, architectures, and checkpoints simultaneously. FreeControl designs structure guidance to facilitate the structure alignment with a guidance image, and appearance guidance to enable the appearance sharing between images generated using the same seed. Extensive qualitative and quantitative experiments demonstrate the superior performance of FreeControl across a variety of…

Wavelet-based ECG compression using dynamic vector quantization with tree codevectors in single codebook

Shaou-Gang Miaou, Heng-Lin Yen, Chih‐Lung Lin

Computer Vision and Pattern Recognition · IEEE Transactions on Biomedical Engineering · July 2002 · cited 112× · DOI ↗ · more like this ·

In this paper, we propose a novel vector quantizer (VQ) in the wavelet domain for the compression of electrocardiogram (ECG) signals. A vector called tree vector (TV) is formed first in a novel structure, where wavelet transformed (WT) coefficients in the vector are arranged in the order of a hierarchical tree. Then, the TVs extracted from various WT subbands are collected in one single codebook. This feature is an advantage over traditional WT-VQ methods, where multiple codebooks are needed and are usually designed separately because numerical ranges of coefficient values in various WT subbands are quite different. Finally, a distortion-constrained codebook replenishment mechanism is incorporated into the VQ, where codevectors can be updated dynamically, to guarantee reliable quality of reconstructed ECG waveforms. With the proposed approach both visual quality and the objective…

Social Isolation, Loneliness, and All-Cause Mortality in Patients With Cardiovascular Disease: A 10-Year Follow-up Study

Bin Yu, Andrew Steptoe, Li‐Jung Chen +3 more

Health · Psychosomatic Medicine · 16 Dec 2019 · cited 110× · DOI ↗ · more like this ·

OBJECTIVE: Social isolation and loneliness have been associated with increased incidence of cardiovascular disease (CVD), but few studies have evaluated the impact of social isolation and loneliness on mortality in people with existing CVD, and these are limited to Western populations. We examined whether social isolation and loneliness are associated with increased risk of mortality in individuals with established CVD in Taiwan. METHODS: The cohort was composed of 1267 patients with confirmed CVD 65 years or older followed up for up to 10 years. Cox proportional hazard regression models were used to examine the association between social isolation and loneliness at baseline and mortality at follow-up by adjusting for demographic variables, health-related behaviors, and health status. RESULTS: There were 593 deaths during the follow-up period. Social isolation was associated with…

Distilling and Retrieving Generalizable Knowledge for Robot Manipulation via Language Corrections

Lihan Zha, Yuchen Cui, Li-Heng Lin +5 more

cs.RO · November 2023 · cited 85× · arXiv ↗ · more like this ·

Today's robot policies exhibit subpar performance when faced with the challenge of generalizing to novel environments. Human corrective feedback is a crucial form of guidance to enable such generalization. However, adapting to and learning from online human corrections is a non-trivial endeavor: not only do robots need to remember human feedback over time to retrieve the right information in new settings and reduce the intervention rate, but also they would need to be able to respond to feedback that can be arbitrary corrections about high-level human preferences to low-level adjustments to skill parameters. In this work, we present Distillation and Retrieval of Online Corrections (DROC), a large language model (LLM)-based system that can respond to arbitrary forms of language feedback, distill generalizable knowledge from corrections, and retrieve relevant past experiences based on…

Mine Your Own vieW: Self-Supervised Learning Through Across-Sample Prediction

Mehdi Azabou, Mohammad Gheshlaghi Azar, Ran Liu +10 more

cs.LG · February 2021 · cited 60× · arXiv ↗ · more like this ·

State-of-the-art methods for self-supervised learning (SSL) build representations by maximizing the similarity between different transformed "views" of a sample. Without sufficient diversity in the transformations used to create views, however, it can be difficult to overcome nuisance variables in the data and build rich representations. This motivates the use of the dataset itself to find similar, yet distinct, samples to serve as views for one another. In this paper, we introduce Mine Your Own vieW (MYOW), a new approach for self-supervised learning that looks within the dataset to define diverse targets for prediction. The idea behind our approach is to actively mine views, finding samples that are neighbors in the representation space of the network, and then predict, from one sample's latent representation, the representation of a nearby sample. After showing the promise of MYOW on…

ShenTu: Processing Multi-Trillion Edge Graphs on Millions of Cores in Seconds

Heng Lin, Xiaowei Zhu, Bowen Yu +9 more

Computer Vision and Pattern Recognition · November 2018 · cited 59× · DOI ↗ · more like this ·

Graphs are an important abstraction used in many scientific fields. With the magnitude of graph-structured data constantly increasing, effective data analytics requires efficient and scalable graph processing systems. Although HPC systems have long been used for scientific computing, people have only recently started to assess their potential for graph processing, a workload with inherent load imbalance, lack of locality, and access irregularity. We propose ShenTu 8 the first general-purpose graph processing framework that can efficiently utilize an entire Petascale system to process multi-trillion edge graphs in seconds. ShenTu embodies four key innovations: hardware specialization, supernode routing, on-chip sorting, and degree-aware messaging, which together enable its unprecedented…

CROSS

Tzu-Heng Lin, Chen Gao, Yong Li

Information Systems · 18 Jul 2019 · cited 57× · DOI ↗ · more like this ·

Social e-commerce, as a new concept of e-commerce, uses social media as a new prevalent platform for online shopping. Users are now able to view, add to cart, and buy products within a single social media app. In this paper, we address the problem of cross-platform recommendation for social e-commerce, i.e., recommending products to users when they are shopping through social media. To the best of our knowledge, this is a new and important problem for all e-commerce companies (e.g. Amazon, Alibaba), but has never been studied before.

The good, the bad and the ugly sides of data augmentation: An implicit spectral regularization perspective

Chi-Heng Lin, Chiraag Kaushik, Eva L. Dyer +1 more

cs.LG · October 2022 · cited 56× · arXiv ↗ · more like this ·

Data augmentation (DA) is a powerful workhorse for bolstering performance in modern machine learning. Specific augmentations like translations and scaling in computer vision are traditionally believed to improve generalization by generating new (artificial) data from the same distribution. However, this traditional viewpoint does not explain the success of prevalent augmentations in modern machine learning (e.g. randomized masking, cutout, mixup), that greatly alter the training data distribution. In this work, we develop a new theoretical framework to characterize the impact of a general class of DA on underparameterized and overparameterized linear model generalization. Our framework reveals that DA induces implicit spectral regularization through a combination of two distinct effects: a) manipulating the relative proportion of eigenvalues of the data covariance matrix in a…

Distributed visual attention on a humanoid robot

Aleš Ude, Valentin Wyart, Li-Heng Lin +1 more

Computer Vision and Pattern Recognition · mediaTUM (Technical University of Munich) · 25 Jan 2006 · cited 49× · DOI ↗ · more like this ·

Complex visual processes such as visual attention are often computationally too expensive to allow real-time implementation on a single computer. To solve this problem we study distributed computer architectures that enable us to divide complex tasks into several smaller problems. In this paper we demonstrate how to implement distributed visual attention system on a humanoid robot to achieve real-time operation at relatively high resolutions and frame rates. We start from a popular theory of bottom-up visual attention that assumes that information across various modalities is used for the early encoding of visual information. Our system uses five different modalities including color, intensity, edges, stereo, and motion. We show how to distribute the attention processing on a computer cluster and study the issues arising on such systems. The system was fully implemented on a workstation…

Quality driven gold washing adaptive vector quantization and its application to ECG data compression

Shaou-Gang Miaou, Heng-Lin Yen

Computer Vision and Pattern Recognition · IEEE Transactions on Biomedical Engineering · 2000 · cited 46× · DOI ↗ · more like this ·

The gold washing (GW) adaptive vector quantization (AVQ) (GW-AVQ) is a relatively new scheme for data compression. The adaptive nature of the algorithm provides the robustness for wide variety of the signals. However, the performance of GW-AVQ is highly dependent on a preset parameter called distortion threshold (dth) which must be determined by experience or trial-and-error. We propose an algorithm that allows us to assign an initial dth arbitrarily and then automatically progress toward a desired dth according to a specified quality criterion, such as the percent of root mean square difference (PRD) for electrocardiogram (ECG) signals. A theoretical foundation of the algorithm is also presented. This algorithm is particularly useful when multiple GW-AVQ codebooks and, thus, multiple dth's are required in a subband coding framework. Four sets of ECG data with entirely different…

Agritourism: experience design, activities, and revisit intention

Austin Rong-Da Liang, Teng-Yuan Hsiao, Dun-Ji Chen +1 more

Sociology and Political Science · Tourism Review · 29 Jun 2020 · cited 45× · DOI ↗ · more like this ·

Purpose Previous studies have discussed individual effects that certain agritourism activities have on visitor response while ignoring interaction effects. Therefore, the purpose of this study is to discuss both the individual and interaction effects of agritourism activities on tourist revisit intention. Design/methodology/approach This study classified agritourism activities into four categories and collected 883 valid samples. Logistic regression was then applied to test the influence of agritourism activities on tourist revisit intention. Findings Based on the results of the statistical analysis, it was found that those agritourism activities, which required mutual cooperation such as do-it-yourself (DIY), animal interaction/feeding and fruit and vegetable picking, enhanced tourist revisit intention. Furthermore, the agritourism groups participating in activities with or without…

Network Analysis of Drug–target Interactions: A Study on FDA-approved New Molecular Entities Between 2000 to 2015

Hui-Heng Lin, Lele Zhang, Ru Yan +2 more

Computational Theory and Mathematics · Scientific Reports · 19 Sep 2017 · cited 45× · DOI ↗ · more like this ·

The U.S. Food and Drug Administration (FDA) approves new drugs every year. Drug targets are some of the most important interactive molecules for drugs, as they have a significant impact on the therapeutic effects of drugs. In this work, we thoroughly analyzed the data of small molecule drugs approved by the U.S. FDA between 2000 and 2015. Specifically, we focused on seven classes of new molecular entity (NME) classified by the anatomic therapeutic chemical (ATC) classification system. They were NMEs and their corresponding targets for the cardiovascular system, respiratory system, nerve system, general anti-infective systemic, genito-urinary system and sex hormones, alimentary tract and metabolisms, and antineoplastic and immunomodulating agents. To study the drug-target interaction on the systems level, we employed network topological analysis and multipartite network projections. As a…

Scalable Graph Traversal on Sunway TaihuLight with Ten Million Cores

Heng Lin, Xiongchao Tang, Bowen Yu +5 more

Computer Vision and Pattern Recognition · May 2017 · cited 44× · DOI ↗ · more like this ·

Interest has recently grown in efficiently analyzing unstructured data such as social network graphs and protein structures. A fundamental graph algorithm for doing such task is the Breadth-First Search (BFS) algorithm, the foundation for many other important graph algorithms such as calculating the shortest path or finding the maximum flow in graphs. In this paper, we share our experience of designing and implementing the BFS algorithm on Sunway TaihuLight, a newly released machine with 40,960 nodes and 10.6 million accelerator cores. It tops the Top500 list of June 2016 with a 93.01 petaflops Linpack performance [1]. Designed for extremely large-scale computation and power efficiency, processors on Sunway TaihuLight employ a unique heterogeneous many-core architecture and memory hierarchy. With its extremely large size, the machine provides both opportunities and challenges for…

Temporal-Spatial Mobile Application Usage Understanding and Popularity Prediction for Edge Caching

Ming Zeng, Tzu-Heng Lin, Min Chen +4 more

Computer Networks and Communications · IEEE Wireless Communications · June 2018 · cited 35× · DOI ↗ · more like this ·

The explosive growth of smart devices and expansion of network services drives the flourish of mobile applications. Caching popular application services at the edge, BSs, or APs, which are closer to end users, could significantly improve the user experience and network capacity. To exploit this potential, it is critical to understand the traffic consumption and app usage pattern under BSs in the metropolitan area. Mobile big data collected from network interfaces facilitates the data-driven approach in characterizing these features. This article aims to design an edge caching strategy for app services based on the observed characteristics of BSs in terms of points of interest (POIs), logs, and traffic generated by various categories of apps. We first analyze the temporal characteristics of different categories of apps, and then further investigate the logs and traffic generated by app…

Cross-platform Item Recommendation for Online Social E-Commerce

Chen Gao, Tzu-Heng Lin, Nian Li +2 more

Information Systems · IEEE Transactions on Knowledge and Data Engineering · 2021 · cited 34× · DOI ↗ · more like this ·

Social e-commerce uses social media as a new prevalent platform for online shopping. In this paper, we address the problem of cross-platform recommendation for social e-commerce, i.e., recommending products to users when they are shopping through social media. To the best of our knowledge, this is a new and important problem for all e-commerce companies (e.g. Amazon, Alibaba), but has never been studied before. Existing cross-platform and social related recommendation methods cannot be applied directly to this problem since they do not co-consider the social information and the cross-platform characteristics together. To study this problem, we collect two real-world datasets from social e-commerce services. We first investigate the heterogeneous shopping behaviors between traditional e-commerce app and social media. Based on these observations from data, we propose CROSS (Cross-platform…

Recommender Systems with Characterized Social Regularization

Tzu-Heng Lin, Chen Gao, Yong Li

cs.IR · 5 Sep 2018 · cited 33× · arXiv ↗ · more like this ·

Social recommendation, which utilizes social relations to enhance recommender systems, has been gaining increasing attention recently with the rapid development of online social network. Existing social recommendation methods are based on the fact that users preference or decision is influenced by their social friends' behaviors. However, they assume that the influences of social relation are always the same, which violates the fact that users are likely to share preference on diverse products with different friends. In this paper, we present a novel CSR (short for Characterized Social Regularization) model by designing a universal regularization term for modeling variable social influence. Our proposed model can be applied to both explicit and implicit iteration. Extensive experiments on a real-world dataset demonstrate that CSR significantly outperforms state-of-the-art social…

Multichannel ECG compression using multichannel adaptive vector quantization

Shaou-Gang Miaou, Heng-Lin Yen

Computer Vision and Pattern Recognition · IEEE Transactions on Biomedical Engineering · 2001 · cited 28× · DOI ↗ · more like this ·

Adaptive vector quantization (AVQ) is a recently proposed approach for electrocardiogram (ECG) compression. The adaptability of the approach can be used to control the quality of reconstructed signals. However, like most of other ECG compression methods, AVQ only deals with the single-channel ECG, and for the multichannel (MC) ECG, coding ECG signals on a channel by channel basis is not efficient, because the correlation across channels is not exploited. To exploit this correlation, an MC version of AVQ is proposed. In the proposed approach, the AVQ index from each channel is collected to form a new input vector. The vector is then vector quantized adaptively using one additional codebook called index codebook. Both the MIT/BIH database and a clinical Holter database are tested. The experimental results show that, for exactly the same quality of reconstructed signals, the MC-AVQ…

Lung Cancer Prediction Using Electronic Claims Records: A Transformer-Based Approach

Huanyu Chen, Huimin Wang, Ching‐Heng Lin +2 more

Artificial Intelligence · IEEE Journal of Biomedical and Health Informatics · 12 Oct 2023 · cited 27× · DOI ↗ · more like this ·

Electronic claims records (ECRs) are large scale and longitudinal collections of individual's medical service seeking actions. Compared to in-hospital medical records (EMRs), ECRs are more standardized and cross-sites. Recently, there has been studies showing promising results on modeling claims data for a wide range of medical applications. However, few of them address the exclusion criteria on cohort selection to extract new incidence without prior signs and also often lack of emphasis on predicting cancer in early stages. In this work, we aim to design a lung cancer prediction framework using ECRs with rigorous exclusion design using state-of-the-art sequence-based transformer. Furthermore, this work presents one of the first results by applying disease prediction model to the entire population in Taiwan. The result shows over 2.1 predictive power, 5 average positive predictive value…

A Market Segmentation Study of Solo Travel Intentions and Constraints

Elaine Chiao Ling Yang, Austin Rong-Da Liang, Jie Heng Lin

Sociology and Political Science · Journal of Hospitality & Tourism Research · 30 Mar 2023 · cited 25× · DOI ↗ · more like this ·

This study provides a market segmentation analysis of the prospective solo travel market based on perceived travel constraints. The data were collected from 1,017 Australian consumers. Using the Fuzzy C-Mean method, the study identified three segments: highly constrained, soloist, and socializer. The segments were distinctive in terms of perceived constraints and solo travel attitudes and intentions. The differences among the segments were investigated further using the concept of self-construal, measured by the level of autonomy and relatedness. Significant differences in sociodemographic characteristics were observed, especially regarding gender, age, and solo dwelling status. Practical recommendations are provided to inform strategies intended to offset travel constraints and create tourism experiences that cater to the needs of different prospective solo travel segments.

Scaling graph traversal to 281 trillion edges with 40 million cores

Huanqi Cao, Yuanwei Wang, Haojie Wang +4 more

Computer Vision and Pattern Recognition · 28 Mar 2022 · cited 25× · DOI ↗ · more like this ·

Graph processing, especially high-performance graph traversal, plays a more and more important role in data analytics. The successor of Sunway TaihuLight, New Sunway, is equipped with nearly 10 PB memory and over 40 million cores, which brings the opportunity to process hundreds of trillions of edges graphs. However, the graph with an unprecedented scale also brings severe performance challenges, including load imbalance, poor locality, and irregular access of graph traversal workload.

Patterns Generation and Spatial Entropy in Two-dimensional Lattice Models

Jung-Chao Ban, Song-Sun Lin, Yin-Heng Lin

Computer Networks and Communications · Asian Journal of Mathematics · 2007 · cited 23× · DOI ↗ · more like this ·

Patterns generation problems in two-dimensional lattice models are studied. Let S be the set of p symbols and Z 22 , 1, be a fixed finite square sublattice of Z 2 . Function U : Z 22 S is called local pattern. Given a basic set B of local patterns, a unique transition matrix A 2 which is a q 2 q 2 matrix, q = p 2 , can be defined. The recursive formulae of higher transition matrix An on Z 2n have already been derived Now A m n , m 1, contains all admissible patterns on Z (m+1)n which can be generated by B. In this paper, the connecting operator Cm, which comprises all admissible patterns on Z (m+1)2 , is carefully arranged. Cm can be used to extend A m n to A m n+1 recursively for n 2. Furthermore, the lower bound of spatial entropy h(A 2 ) can be derived through the diagonal part of Cm. This yields a powerful method for verifying the positivity of spatial entropy which is important in…

Item Recommendation for Word-of-Mouth Scenario in Social E-Commerce

Chen Gao, Chao Huang, Donghan Yu +4 more

Information Systems · IEEE Transactions on Knowledge and Data Engineering · 20 Aug 2020 · cited 16× · DOI ↗ · more like this ·

Social commerce, which is different from traditional e-commerce where people purchase products via initiative searching or recommendations from the platform, transforms a social community into an inclusive place to do business by enabling people to share products with their friends. A user ( sharer ), can share a link of a product to their social-connected friends ( receiver ). Once a receiver purchases the product, the sharer can earn commission provided by the platform. To promote sales, the platform can also assist sharers by providing product candidates which are more likely to be purchased during the social sharing. We define this task of generating sharing suggestions as item recommendation for word-of-mouth scenario, and to the best of our knowledge, this is a new task that has never been explored. In this article, we propose a TriM (short for Tri ad…

Social Recommendation with Characterized Regularization

Chen Gao, Nian Li, Tzu-Heng Lin +4 more

Information Systems · IEEE Transactions on Knowledge and Data Engineering · 2020 · cited 15× · DOI ↗ · more like this ·

Social recommendation, which utilizes social relations to enhance recommender systems, has been gaining increasing attention recently with the rapid development of online social networks. Existing social recommendation methods are based on the assumption, so-called <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">social-trust , that users’ preference or decision is influenced by their social-connected friends’ purchase behaviors. However, they assume that the influences of social relationships are always the same, which violates the fact that users are likely to share preference on different products with different friends. More precisely, friends’ behaviors do not necessarily affect a user’s preferences, and the influence is diverse among different items. In this paper, we contribute a new solution, CSR (short for <bold…

On the Understanding of Video Streaming Viewing Behaviors Across Different Content Providers

Huan Yan, Tzu-Heng Lin, Chuhan Gao +2 more

Computer Networks and Communications · IEEE Transactions on Network and Service Management · 23 Jan 2018 · cited 15× · DOI ↗ · more like this ·

Watching videos online has become prevalent in our daily activities. While there are extensive measurements studying about video on demand, IPTV, live streaming, etc., most of them focus their analysis on a single content provider (CP). Various user behaviors over multiple different CPs are yet to be explored. In this paper, we collect about ten million viewing records of the six most popular video CPs through a major Internet service provider (ISP) in Shanghai, China, and examine the user behaviors and access patterns with an emphasis on comparing them among different CPs in the same environments. Our analysis reveals the day-scale temporal patterns, the diversified device-wise preferences, and the city-scale geographical features of the different CPs, and identifies the high relevance between the user behaviors and the contents of the videos. Moreover, we find that user migration…

Domestic violence against pregnant women is a potential risk factor for low birthweight in full-term neonates: A population-based retrospective cohort study

Ching‐Heng Lin, Wei‐Szu Lin, Hsiu-Yuan Chang +1 more

Health · PLoS ONE · 22 Dec 2022 · cited 14× · DOI ↗ · more like this ·

Domestic violence's most frequently reported outcomes are preterm delivery and low birthweight, both of which are the strongest correlates of mortality and morbidity. Several studies have shown that pregnant women with domestic violence during pregnancy were more likely to deliver low-birthweight and preterm neonates. However, there has been no consensus on associations between domestic violence and low-birthweight and preterm delivery. To examine the impact of domestic violence on birthweight stratified by preterm or full-term delivery, a population-based retrospective cohort study was conducted that linked four national databases in Taiwan. A total of 1,322 subjects associated with a report of domestic violence during pregnancy were compared with 485,981 subjects without any record of reported domestic violence. The percentage of low birthweight in the group exposed to domestic…