About

Professor Mingyan Liu is a distinguished researcher with a focus on memory devices and memory-centric computing systems, including resistive random-access memory (RRAM) and memristors, as well as neuromorphic circuits and systems. Her research interests extend to novel transistor devices, neuro-inspired AI models, and compute-in-memory architectures. She has made significant contributions to the development and understanding of resistive switching mechanisms, memristor networks, and their applications in real-time neural activity analysis, temporal data classification, and advanced in-memory computing systems. Mingyan Liu's academic background includes a B.S. degree from Tsinghua University and a Ph.D. from Rice University. Her research career has involved extensive work on the physical modeling of resistive switching, the design of memristor-based neural networks, and the implementation of innovative computing architectures. She has authored numerous publications in high-impact journals and conferences, and her work has been featured in prominent scientific outlets. Her contributions have advanced the field of nanoelectronics and neuromorphic computing, and she is recognized for her leadership in developing memristor technologies for AI and data processing applications.

Research topics

• Computer Science
• Computer Security
• Artificial Intelligence
• Mathematics
• Machine Learning
• Management science
• Operations research
• Engineering
• Psychology
• Remote sensing
• Mathematical optimization
• Computer vision

Selected publications

• Estimating the Social Cost of Corporate Data Breaches

  ArXiv.org · 2026-03-22

  articleOpen accessSenior author

  While the size of a data breach is typically measured by the number of (consumer, customer, or user) records exposed or compromised, its economic impact is generally measured from the point of view of the corporation suffering the data breach: cost in crisis management, legal fees, drop in stock price, and so on. This study examines whether it is possible to estimate the true cost, or the social cost of a data breach, measured by the impact on its victims and their out of pocket costs. To accomplish this we establish: (1) the estimation of the average direct financial losses of an identity theft (IDT) victim, including the opportunity cost of lost time, and healthcare expenditures associated with distress associated with identity theft; and (2) the estimation of increases in incidents of IDT that can be attributed to a major breach event. Our findings show that the average social cost per victim has declined significantly since 2016. Furthermore, we find that there is indeed a statistically significant increase in the number of IDTs following a mega-breach event when accounting for a discovery lag of 1-2 months post-breach. Applying our model to real-world cases allows us to estimate an upper and lower bound social cost of specific mega-breach events. We find that for the 2009 Heartland and 2013 Target breaches, even the conservative lower bound social cost estimate exceeded settlements by factors of 5 and 18, respectively. In contrast, the 2017 Equifax breach resulted in a lower bound estimate of $263.8 million, falling well within its $700 million settlement cap. While the Equifax upper bound estimate of $1.72 billion in social cost more than doubles this settlement, the narrowing gap between institutional liability and an incident's social cost provides empirical evidence of a market saturation effect that reduces the marginal damage of individual compromised records over time.

  PublisherOA PDF

• Research on Fracturing Sweet Spot and Cluster Design Based on Big Data Analysis and Intelligent Fusion Algorithms

  Chemistry and Technology of Fuels and Oils · 2025-03-01 · 1 citations

  article
  PublisherDOI

• How family support alleviates death anxiety in breast cancer patients: the mediating role of meaning in life

  Frontiers in Public Health · 2025-04-01 · 14 citations

  articleOpen access

  Previous studies have predominantly focused on the relationship between death anxiety and quality of life in breast cancer patients, with limited exploration on how to alleviate their death anxiety. To address this gap, we recruited 533 breast cancer patients and utilized structural equation modeling and Process Model 4 to analyze the internal mechanisms and boundary conditions between family support and death anxiety. The study results indicated that family support significantly negatively impacts death anxiety in breast cancer patients; similarly, meaning in life also significantly negatively impacts death anxiety. More importantly, we found that meaning in life plays a full mediating role between family support and death anxiety. This study suggests that by enhancing family support levels and strengthening patients' perception of meaning in life, we can significantly improve the psychological health status of breast cancer patients, thereby potentially improving their quality of life.

  PublisherOA PDFDOI

• Scoring the Unscorables: Cyber Risk Assessment Beyond Internet Scans

  ArXiv.org · 2025-06-07

  preprintOpen accessSenior author

  In this paper we present a study on using novel data types to perform cyber risk quantification by estimating the likelihood of a data breach. We demonstrate that it is feasible to build a highly accurate cyber risk assessment model using public and readily available technology signatures obtained from crawling an organization's website. This approach overcomes the limitations of previous similar approaches that relied on large-scale IP address based scanning data, which suffers from incomplete/missing IP address mappings as well as the lack of such data for large numbers of small and medium-sized organizations (SMEs). In comparison to scan data, technology digital signature data is more readily available for millions of SMEs. Our study shows that there is a strong relationship between these technology signatures and an organization's cybersecurity posture. In cross-validating our model using different cyber incident datasets, we also highlight the key differences between ransomware attack victims and the larger population of cyber incident and data breach victims.

  PublisherOA PDFDOI

• Feedforward Control and Construction of Control System of Postpartum Hemorrhage in Primary Hospital.

  PubMed · 2025-10-01

  article

  Objective: To explore the effect of feedforward control and control system on reducing postpartum hemorrhage (PPH) rate in primary hospitals. Methods: Multi-period PPH special training classes were held in five primary hospitals, such as the Ruyuan Yao Autonomous County People's Hospital, Wengyuan County People's Hospital, located in Shaoguan, Guangdong, China, from 2017 to 2018. In this study, theoretical teaching combined with practical operations were performed. Clinical case-control analysis of PPH risk before and after PPH training was carried out. Results: After training, the overall incidence of PPH decreased by 0.44% in the five hospitals (χ2 = 7.16, P = .007), the incidence of PPH in natural delivery decreased by 0.47% (χ2 = 5.41, P = .02), and decreased by 0.37% in cesarean section (χ2 = 1.86, P = .173). Uterine atony was the most common cause of PPH, but placental factors caused greater amount of delivery hemorrhage and bleeding 24 hours after delivery in the cases of PPH than uterine atony. Conclusion: Adopting feedforward control management and constructing a control system in Shaoguan primary hospitals could effectively reduce the incidence of PPH. It is imperative to strengthen the training of pregnancy management and hemostasis surgery for pregnant women with placental factors. Keywords: PPH, primary hospitals, feedforward control, control system.

  Publisher

• Secure Ranging for Proximity-Based Authentication Using Physical Unclonable Functions in OFDM

  2025-12-08

  article

  This paper presents a secure ranging system for proximity-based authentication using orthogonal frequency division multiplexing (OFDM) and Physically Unclonable Function (PUF). It specifically targets applications such as keyless entry systems that require accurate and tamper-resistant distance measurement. The proposed system utilizes PUF-based noiselike signatures embedded in the OFDM modulation process to mitigate distance manipulation attacks from adversaries. We implemented the design on a USRP X310 platform, leveraging both software and hardware-based processing to achieve real-time performance. Experimental evaluations conducted in wireless channels demonstrate that our PUF-based secure ranging system achieves decimeter-level accuracy and successfully resists distance reduction attacks. This approach offers high security and precision, making it a promising solution for secure and robust proximity-based authentication.

  PublisherDOI

• A Unified Virtual Mixture-of-Experts Framework:Enhanced Inference and Hallucination Mitigation in Single-Model System

  ArXiv.org · 2025-04-01

  preprintOpen access1st authorCorresponding

  Generative models, such as GPT and BERT, have significantly improved performance in tasks like text generation and summarization. However, hallucinations "where models generate non-factual or misleading content" are especially problematic in smaller-scale architectures, limiting their real-world applicability.In this paper, we propose a unified Virtual Mixture-of-Experts (MoE) fusion strategy that enhances inference performance and mitigates hallucinations in a single Qwen 1.5 0.5B model without increasing the parameter count. Our method leverages multiple domain-specific expert prompts (with the number of experts being adjustable) to guide the model from different perspectives. We apply a statistical outlier truncation strategy based on the mean and standard deviation to filter out abnormally high probability predictions, and we inject noise into the embedding space to promote output diversity. To clearly assess the contribution of each module, we adopt a fixed voting mechanism rather than a dynamic gating network, thereby avoiding additional confounding factors. We provide detailed theoretical derivations from both statistical and ensemble learning perspectives to demonstrate how our method reduces output variance and suppresses hallucinations. Extensive ablation experiments on dialogue generation tasks show that our approach significantly improves inference accuracy and robustness in small models. Additionally, we discuss methods for evaluating the orthogonality of virtual experts and outline the potential for future work involving dynamic expert weight allocation using gating networks.

  PublisherOA PDFDOI

• Association of UHR and ECG parameters with type 2 diabetes mellitus in non-alcoholic fatty liver disease

  Frontiers in Endocrinology · 2025-08-28

  articleOpen access

  Background: Non-alcoholic fatty liver disease (NAFLD) is the most prevalent liver disease globally. NAFLD increases the risk of type 2 diabetes mellitus (T2DM) while lacking clinical predictors. This study aims to investigate the characteristics and clinical significance of the uric acid (UA) to high-density lipoprotein cholesterol ratio (UHR) and electrocardiogram (ECG) parameters in NAFLD patients, both with and without T2DM. Methods: We compared 102 NAFLD with T2DM (NAFLD-T2DM) cases to 113 NAFLD without T2DM (NAFLD-nT2DM) cases. Baseline data and biochemical indicators, including UHR, were collected and analyzed in each group. A 12-lead ECG was used to collect parameters that were compared between the two groups. Multivariate logistic regression analysis was employed to identify factors influencing NAFLD with T2DM. Receiver operating characteristic (ROC) curves were utilized to assess the clinical value of UHR combined with ECG parameters in identifying T2DM risk among NAFLD patients. Results: Compared to the NAFLD-nT2DM group, the NAFLD-T2DM group exhibited significantly higher levels of triglycerides (TG), fasting plasma glucose (FPG), UA, and UHR, while aspartate aminotransferase (AST) levels were lower (P < 0.05). The incidence of ST-T changes, heart rate, and P wave duration was also higher in the NAFLD-T2DM group, whereas the QT interval was shorter (P < 0.05). Multivariate logistic regression analysis revealed that UHR, ST-T changes, heart rate, QT interval, and P wave duration are independent factors associated with the incidence of T2DM in NAFLD. ROC curve analysis indicated that the area under the curve (AUC) for the combination of five variables in predicting T2DM in NAFLD was 0.949 (95% CI: 0.905-0.977, P < 0.05), with a sensitivity of 91.96% and a specificity of 93.55%, significantly superior to those of individual indicators. Conclusion: UHR and ECG parameters are associated with T2DM in NAFLD patients. The combination of UHR and ECG parameters demonstrates predictive value for the incidence of T2DM in NAFLD patients. Clinical attention should be directed toward the levels of UHR and ECG parameters in NAFLD with T2DM.

  PublisherOA PDFDOI

• Comparing Object Detection, Instance Segmentation, and Semantic Segmentation for Automated Vegetation Detection in Railroad Systems

  2025-10-22

  article1st authorCorresponding

  Vegetation management is a critical component of railway maintenance, directly impacting operational safety, infrastructure longevity, and regulatory compliance. Overgrown vegetation can obscure track visibility, interfere with inspection routines, and degrade track conditions which posing risks to both freight and passenger operations. This paper explores automated vegetation detection within railroad environments using three deep learning models: YOLOv8, U-Net, and DeepLabv3+. The YOLOv8 model was trained using both object detection bounding boxes and instance segmentation masks on a domain-specific dataset comprising around 500 railroad images captured under real-world deployment conditions. In contrast, the semantic segmentation models, U-Net and DeepLabv3+, were trained on a broader dataset of more than 9,800 images representing general vegetation contexts. Comparative analysis reveals that DeepLabv3+ consistently outperforms the other models in accurately identifying vegetation, demonstrating higher precision, recall, and segmentation quality. These findings highlight the effectiveness of semantic segmentation, particularly DeepLabv3+, for detecting irregular, organic features such as vegetation in complex railway settings.

  PublisherDOI

• Study on the properties of high-performance semi-flowable self-compacting concrete suitable for road works

  Construction and Building Materials · 2025-11-20 · 4 citations

  article
  PublisherDOI

Recent grants

• NeTS: Small: Collaborative Research: Playing the Devil's Advocate: The Profit Perspective in Secondary Spectrum Markets

  NSF · $215k · 2012–2015

• TTP: Small: Network-Level Security Posture Assessment and Predictive Analytics: From Theory to Practice

  NSF · $516k · 2016–2021

• EAGER: Theory and Practice of Risk-Informed Cyber Insurance Policies: Risk Dependency, Risk Aggregation, and Active Threat Landscape

  NSF · $216k · 2019–2022

• CAREER: Capacity-Driven Design of Large-Scale Wireless Sensor Networks

  NSF · $422k · 2003–2010

• TWC: Small: Understanding Network Level Malicious Activities: Classification, Community Detection and Inference of Security Interdependence

  NSF · $500k · 2014–2018

Frequent coauthors

• Parinaz Naghizadeh

  32 shared

• Chaowei Xiao

  30 shared

• Cem Tekin

  24 shared

• Mohammad Mahdi Khalili

  23 shared

• Xueru Zhang

  17 shared

• Jon Crowcroft

  University of Cambridge

  17 shared

• Qingsi Wang

  Sichuan Academy of Medical Sciences & Sichuan Provincial People's Hospital

  16 shared

• Kun Jin

  University of Michigan–Ann Arbor

  14 shared

Labs

• Wei Lu Research GroupPI

Awards & honors

• IEEE Fellow
• AAAS Fellow
• NSF CAREER Award
• 2012 EECS Outstanding Achievement Award
• 2014-15 Rexford E. Hall Innovation Excellence Award