About

Chen Li is a professor in the Department of Computer Science at UC Irvine, specializing in data management, including data-intensive computing, databases, query processing and optimization, machine learning-based systems, data analytics, search, and visualization. He received his Ph.D. in Computer Science from Stanford University in 2001, and his M.S. and B.S. in Computer Science from Tsinghua University, China. Chen Li has been recognized with an NSF CAREER award and several test-of-time publication awards. He has served as a part-time visiting research scientist at Google, was a PC co-chair of VLDB 2015, and is an ACM distinguished member and an IEEE fellow. Since January 2020, he has been the treasurer and a board member of the VLDB Endowment, and since July 2020, he has served as the Faculty Director of the ICS Master of Computer Science Program. He was also a co-founder and CTO of a startup to commercialize his research results.

Research topics

• Computer Science
• Artificial Intelligence
• Psychology
• Political Science
• Sociology
• World Wide Web
• Algorithm
• Medicine
• Cognitive psychology
• Distributed computing
• Speech recognition
• Internet privacy
• Computer network
• Operating system
• Neuroscience
• Pathology
• Nursing

Selected publications

• Strategies for Enhancing the Adaptability of Vocational Education IoT Application Technology Talents from the Perspective of Game Theory

  Education and Social Work · 2025-03-25 · 1 citations

  articleOpen access1st authorCorresponding

  The explosive growth of the Internet of Things industry and the structural con-tradiction in talent cultivation have given rise to a complex game relationship among schools, enterprises, and students. This article constructs a three party dynamic evolutionary game model, revealing core game factors such as imbal-anced distribution of interests, institutional design flaws, conflicts in subject roles, and information asymmetry in school enterprise cooperation. Research has shown that net profit from enterprise participation is a key variable driving cooperation, and student rights protection and third-party evaluation mecha-nisms can effectively improve the efficiency of game equilibrium. Through the three in one reform framework of "policy incentives collaborative mechanisms subject empowerment", measures such as differentiated subsidies, industry de-mand warning platforms, and flexible credit systems are proposed to promote the game from zero sum competition to symbiotic value-added.

  PublisherDOI

• Adaptive feature-extraction graph network for physical systems: Prediction of inviscid compressible flow in urban explosion

  Engineering Structures · 2025-09-30 · 1 citations

  articleSenior author
  PublisherDOI

• Harnessing CO2 fixation and reducing power recycling for enhanced polyhydroxyalkanoates industrial bioproduction

  Metabolic Engineering · 2025-05-01 · 12 citations

  article
  PublisherDOI

• The whole-brain structural and functional connectome in Alzheimer’s disease spectrum: A multimodal Bayesian meta-analysis of graph theoretical characteristics

  Neuroscience & Biobehavioral Reviews · 2025-04-24 · 6 citations

  reviewOpen access

  Alzheimer’s disease (AD) spectrum is increasingly recognized as a progressive network-disconnection syndrome. Neuroimaging studies using graph theoretical analysis (GTA) have reported alterations in the topological properties of whole-brain structural and functional connectomes in both preclinical AD and AD patients, though findings remain inconsistent. This study aimed to identify robust changes in multimodal GTA metrics across the AD spectrum through a comprehensive literature search and Bayesian random-effects meta-analyses. The analysis included 53 studies (37 functional and 17 structural), involving 1649 AD patients, 1455 preclinical AD patients, and 1771 healthy controls (HC). Results revealed lower structural network integration (evidenced by higher characteristic path length and/or normalized characteristic path length) and segregation (evidenced by lower clustering coefficient and local efficiency) in AD and preclinical AD patients compared to HC. Functional network segregation was also lower in AD patients, while preclinical AD showed preserved functional topology despite structural changes. Moderator analyses identified potential methodological moderators, including neuroimaging technique, node and edge definitions, and network type, although further validation is needed. These findings support the progressive disconnection hypothesis in the AD spectrum and suggest that structural network alterations may precede functional network changes. Furthermore, the results help clarify inconsistencies in previous studies and highlight the utility of graph-based metrics as biomarkers for staging AD progression. • A meta-analysis included 53 studies with 1649 Alzheimer’s patients (AD), 1455 preclinical AD patients, and 1771 controls. • Structural and functional network change in AD, while preclinical AD shows structural changes with preserved functional topology. • Brain network alterations emerge in the preclinical stage of AD and worsens with disease progression. • Neuroimaging techniques, node/edge definitions, and network types may moderate outcomes, but further validation is needed.

  PublisherDOI

• Evaluation of diffusion MRI-based cytometry imagine based on Spearman's correlation analysis

  Proceedings on CD-ROM - International Society for Magnetic Resonance in Medicine. Scientific Meeting and Exhibition/Proceedings of the International Society for Magnetic Resonance in Medicine, Scientific Meeting and Exhibition · 2025-09-16

  article

  Motivation: DMRI-based cytometry methods have emerged for in-vivo microstructural imaging, however, it is difficult to evaluate accuracy of fitted results in clinic due to lack of ground-truth. Goal(s): To quantitative evaluate accuracy of results through correlation analysis. Approach: IMPULSED fitting and Spearman's correlation analysis on dMRI data from 90 patients with malignant breast cancer. numerical simulations Results: Strong correlation metrics between IMPULSED-derived microstructural parameters and measured ADC values can be used to evaluate the accuracy of model fitting, such as the metric of νin vs. ADCPGSE , which is higher to indicate lower fitted errors. Impact: Evaluation for accuracy of emerging dMRI-based cytometry methods is limited by the lack of ground-truth, such as pathologic staining results. This prospective study demonstrated clinical potential of evaluating accuracy of imaging results through correlation analysis in the absence of ground-truth.

  PublisherDOI

• Image Captioning via Masked Conditional Diffusion

  Lecture notes in computer science · 2025-10-18

  book-chapter
  PublisherDOI

• Laser-Processed All-Paper-Based Wearable and Biodegradable Electrochemical Sensor for Continuous Sweat pH Detection

  ACS Sustainable Chemistry & Engineering · 2025-10-13 · 2 citations

  article1st author

  Flexible and wearable electronics have attracted widespread attention in human healthcare and medical applications owing to their inherent flexibility, skin conformability, and continuous real-time operating capabilities. However, conventional petroleum-based polymer wearable electronics lack environmental sustainability as their disposal generates hazardous e-waste with significant ecological impacts. Herein, we fabricated a biodegradable and wearable electrochemical sensor with typical cellulose-based paper for continuous monitoring of sweat pH. The paper was hydrophobically modified, and etched by CO2 laser, then layer–layer assembled to integrate an all-paper-based electrochemical (APEC) sensor. The APEC sensor consists a paper-based pH sensor and a microfluidic patch for sweat sampling. We constructed MXene-supported polyaniline:poly(styrenesulfonate) (PANI:PSS) nanoparticles for detecting H+, which are water-soluble because of the introduction of PSS, rendering them compatible for fabricating paper-based sensors. The APEC sensors demonstrate biodegradability in both aqueous and soil media. Their waterproof characteristics enable reliable operation during in situ sweat sampling and real-time sweat analysis. These sensors exhibit an ultrawide linear pH detection range (pH 1.53–13.46) with excellent reversibility and high sensitivity (46.01 mV pH–1). On-body measurements demonstrated that the APEC sensors can continuously detect the pH of human sweat and have the potential to detect muscle fatigue. The fabrication and demonstration of the APEC sensors is expected to facilitate the development of flexible and wearable electronics and contribute to a sustainable electronics future.

  PublisherDOI

• PipC affects the virulence of Salmonella enterica serovar Enteritidis and its deletion strain provides effective immune protection in mice

  Frontiers in Microbiology · 2025-06-24

  articleOpen access

  Background Salmonellosis caused by Salmonella sp. is a foodborne zoonotic disease that poses a significant threat to public health security. Vaccination is a safe and effective strategy for preventing and controlling Salmonella infections. PipC is a chaperone protein associated with Salmonella invasion proteins which is crucial for bacteria to invade host cells. Methods In this study, a Δ pipC mutant strain was generated. Subsequently, we examined the environmental stress tolerance of the mutant strain through in vitro simulation experiments. Moreover, its virulence by employing cell and mouse infection models was investigated. Furthermore, we utilized a mouse model to further explore its potential as an attenuated live vaccine against Salmonella enterica serovar Enteritidis infection. Results The Salmonella strain C50336 with a deletion of the pipC gene exhibits a significant reduction in its ability to resist environmental stress and virulence. Meanwhile, the expression levels of SPI-1-related genes ( invH , sipA , sipB , sipC , sopB , and sopE2 ) and SPI-2-related genes ( spvB , ssrA , orf245 , ssaS , ssaT , ssaU , sseB , and sseD ) encoding the Salmonella type III secretion system (T3SS) were found to be decreased, leading to a significant reduction in the bacteria’s invasion and intracellular survival abilities. The results of the mouse intraperitoneal challenge experiment showed that compared with the wild-type strain, the 50% lethal dose (LD 50 ) of the Δ pipC strain increased by 47 times, and the bacterial loads in the liver, spleen, and cecum were significantly reduced. When mice were immunized with the Δ pipC mutant strain, the immunized mice showed a robust immune response, with significantly increased cytokine and antibody levels in their bodies. Mice vaccinated with the Δ pipC mutant strain had 100% immune protection against wild-type Salmonella infection. Conclusion This study demonstrates that lack of pipC affects SE pathogenicity by decreasing its virulence both in vitro and in vivo . Vaccination of mice with Δ pipC conferred development of an acquired immunity and efficacious protection against experimental systemic infection. These results indicated that the Δ pipC mutant strain can be used in the development of attenuated live vaccines.

  PublisherOA PDFDOI

• Intelligent layout of nuclear pipeline system with mechanical constraint

  Progress in Nuclear Energy · 2025-09-18

  article
  PublisherDOI

• InstGAN: Instant Actor-Critic-Driven GAN for De Novo Molecule Generation and Property Optimization

  2025-09-01

  article

  Deep generative models, such as generative adversarial networks (GANs), have been employed for de~novo molecular generation in drug discovery. Most prior studies have utilized reinforcement learning (RL) algorithms, particularly Monte Carlo tree search (MCTS), to handle the discrete nature of molecular representations in GANs. However, due to the inherent instability in training GANs and RL models, along with the high computational cost associated with MCTS sampling, MCTS RL-based GANs struggle to scale to large chemical databases. To tackle these challenges, this study introduces a novel GAN based on actor-critic RL with instant and global rewards, called InstGAN, to generate molecules at the token-level with multi-property optimization. Furthermore, maximized information entropy is leveraged to alleviate the mode collapse. The experimental results demonstrate that InstGAN outperforms other baselines, achieves comparable performance to state-of-the-art models, and efficiently generates molecules with multi-property optimization. The code is available at: https://github.com/tang777777/InstGAN.

  PublisherDOI

Recent grants

• EAGER: Supporting GUI-Based Text Analytics on Social Media Data by Non-Technical Users

  NSF · $216k · 2017–2019

• CAREER: Peer-Based Data Integration and Sharing of Heterogeneous Sources

  NSF · $400k · 2003–2008

• Supporting Efficient Fuzzy Queries on Large Text Repositories Using Hadoop

  NSF · $222k · 2009–2012

Frequent coauthors

• Junjun Zheng

  Hiroshima University

  20 shared

• Hiroyuki Okamura

  Hiroshima University

  19 shared

• Tadashi Dohi

  19 shared

• Michael J. Carey

  University of California, Irvine

  18 shared

• Yasuhiko Morimoto

  16 shared

• Alexander Behm

  Databricks (United States)

  14 shared

• Yoshihiro Yamanishi

  Kyushu Institute of Technology

  12 shared

• Guoliang Li

  Tsinghua University

  12 shared

Education

• PhD, Graduate School of Advanced Science and Engineering

  Hiroshima University

  2019

• Master, Graduate School of Advanced Science and Engineering

  Hiroshima University

  2019

Awards & honors

• NSF CAREER award
• test-of-time publication awards
• ACM distinguished member
• IEEE fellow
• IEEE Fellow for Significant Contributions to Approximate Que…