About

Wenqi "Marcus" Wu is a PhD student in the Department of Communication at the University of Maryland. His research interests include Public Relations and Crisis Communication. Marcus earned his bachelor's degree in communication and sociology from the University of Connecticut. His academic background and research focus are centered on understanding communication strategies in crisis and risk scenarios, contributing to the fields of public relations and communication studies.

Research topics

• Computer Science
• Artificial Intelligence
• Nursing
• Process management
• Medicine
• Business

Selected publications

• Intelligent Diagnosis and Fault Location in Distribution Transformer Areas with Hybrid Access of Diverse Loads

  2026-03-20

  article

  In response to the complexity introduced by the large-scale integration of diverse loads into the operational state of distribution transformer areas, as well as the inadequacy of passive monitoring modes in effectively providing early fault warnings and locating risks, this study investigates intelligent diagnosis and fault location methods for distribution transformer areas. Empirical Mode Decomposition (EMD) technology is employed to separate high- and low-frequency components representing different source-load characteristics from load curves. A multi-objective optimization diagnostic model integrating multi-source information and protection logic is constructed. Additionally, a fault section location method based on voltage clustering analysis and topological reasoning is incorporated, forming a closed-loop active defense framework that spans from detailed analysis of load characteristics and intelligent assessment of operational states to precise fault area localization. Case studies demonstrate that the proposed method can effectively decompose hybrid load characteristics, accurately identify composite operational risks, and achieve section-level precise location in simulated fault scenarios.

  PublisherDOI

• Phonetically-Augmented Discriminative Rescoring for Voice Search Error Correction

  2025-08-17

  article
  PublisherDOI

• Phonetically-Augmented Discriminative Rescoring for Voice Search Error Correction

  ArXiv.org · 2025-06-06

  preprintOpen access

  End-to-end (E2E) Automatic Speech Recognition (ASR) models are trained using paired audio-text samples that are expensive to obtain, since high-quality ground-truth data requires human annotators. Voice search applications, such as digital media players, leverage ASR to allow users to search by voice as opposed to an on-screen keyboard. However, recent or infrequent movie titles may not be sufficiently represented in the E2E ASR system's training data, and hence, may suffer poor recognition. In this paper, we propose a phonetic correction system that consists of (a) a phonetic search based on the ASR model's output that generates phonetic alternatives that may not be considered by the E2E system, and (b) a rescorer component that combines the ASR model recognition and the phonetic alternatives, and select a final system output. We find that our approach improves word error rate between 4.4 and 7.6% relative on benchmarks of popular movie titles over a series of competitive baselines.

  PublisherOA PDFDOI

• Analysis of ENF Signal Extraction From Videos Acquired by Rolling Shutters

  2025-06-15

  articleOpen accessSenior author

  Electric network frequency (ENF) analysis is a promising forensic technique for authenticating multimedia recordings and detecting tampering. The validity of the ENF analysis heavily relies on the capability of extracting high-quality ENF signals from multimedia recordings. This paper analyzes and compares two representative methods for extracting ENF signals from visual signals acquired by cameras using the rollingshutter mechanism. The first method proposed in prior work, direct concatenation, ignores the idle period of each frame. The second method proposed in this paper, periodic zeroing-out, inserts zeros to missing sample points instead of ignoring the idle period. Our theoretical analyses of using multirate signal processing reveal and experiments confirm that while the first method can extract ENF signals without knowing the exact value of camera read-out time, there exists some mild distortion to extracted ENF signals. In contrast, the second method taking the read-out time as the additional input is capable of extracting distortion-free ENF signals, and its frequency component of the highest strength is always located at the nominal frequency. Additionally, we examine aliased DC and negative ENF components caused by the two methods and show that their impact on the accuracy of frequency estimation is minimal. This paper facilitates the fundamental understanding of extracting ENF signals from videos. The research findings imply that the periodic zeroing-out method offers more accurate frequency estimates, but the performance improvement is moderate.

  PublisherOA PDFDOI

• Feasibility Study of Location Authentication for IoT Data Using Power Grid Signatures

  IEEE Open Journal of Signal Processing · 2025-01-01 · 2 citations

  articleOpen accessSenior author

  Ambient signatures related to the power grid offer an under-utilized opportunity to verify the time and location of sensing data collected by the Internet-of-Things (IoT). Such power signatures as the Electrical Network Frequency (ENF) have been used in multimedia forensics to answer questions about the time and location of audio-visual recordings. Going beyond multimedia data, this paper investigates a refined power signature of Electrical Network Voltage (ENV) for IoT sensing data and carries out a feasibility study of location verification for IoT data. ENV reflects the variations of the power system's supply voltage over time and is also present in the optical sensing data, akin to ENF. A physical model showing the presence of ENV in the optical sensing data is presented along with the corresponding signal processing mechanisms to estimate and utilize ENV signals from the power and optical sensing data as location stamps. Experiments are conducted in the State of Maryland of the United States to demonstrate the feasibility of using ENV signals for location authentication of IoT data.

  PublisherDOI

• Exploring Nonlinear Threshold Effects and Interactions Between Built Environment and Urban Vitality at the Block Level Using Machine Learning

  Land · 2025-06-07 · 9 citations

  articleOpen access

  Urban vitality is a critical indicator of both urban sustainability and quality of life. However, comprehensive studies examining the threshold effects and interaction mechanisms of built environment factors on urban vitality at the block level remain limited. This study proposed to develop a comprehensive framework for urban vitality by incorporating multi-source data, and the central urban area of Xi’an, China, was selected as the study area. Four machine learning models, LightGBM, XGBoost, GBDT, and random forest, were employed to identify the most fitted model for analyzing threshold effects and interactions among built environment factors on shaping urban vitality. The results showed the following: (1) Xi’an’s urban vitality exhibited a distinct gradient, with the highest vitality concentrated in the Yanta District; (2) life service facility density was the most significant determinant of vitality (19.91%), followed by air quality (9.01%) and functional diversity (6.49%); and (3) significant interactions among built environment factors were observed. In particular, streets characterized by both high POI diversity (greater than 0.8) and low PM2.5 concentrations (below 48.5 μg/m3) exhibited notably enhanced vitality scores. The findings of this study provide key insights into strategies for boosting urban vitality, offering actionable insights for improving land use allocations and enhancing quality of life.

  PublisherOA PDFDOI

• Two-Bit RIS-Aided NLOS DOA Based on Fast Zoomed SBL

  IEEE Transactions on Aerospace and Electronic Systems · 2025-10-13

  article1st authorCorresponding

  Direction of arrival (DOA) estimation of non-line-of-sight (NLOS) links assisted by reconfigurable intelligent surface (RIS) is a critical technology in the wireless communication field. The phase control precision and implementation complexity of RIS systems exhibit direct dependence on the quantization bit depth employed in their configuration. Furthermore, the estimation accuracy of the RIS aided DOA estimation methodology under the SBL framework experiences substantial degradation when grid mismatch conditions occur. In this article, we design a two-bit RIS aided system, and propose a DOA estimation method based on fast zoomed sparse Bayesian learning (SBL). Firstly, under the minimum weighted integral sidelobe level (WISL) criterion, a coding scheme of two-bit orthogonal phase matrix is proposed to realize low cost and accurate phase control by using four adjustable phases. Secondly, we construct an observation model composed of Fourier matrices, and then propose a zoomed dictionary based SBL algorithm. This approach yields more accurate angular information. Additionally, a circulant-skew circulant Gohberg-Semencul (CSCGS) decomposition algorithm is proposed to calculate the inversion of covariance matrix in the iterative process of SBL, which can significantly reduce the computational complexity. Simulation results demonstrate the feasibility of the proposed method under any number of snapshots and its superior performance compared to existing methods.

  PublisherDOI

• Effects of therapeutic instrumental music performance on upper limb motor function of children with cerebral palsy: A systematic review

  Brain and Development · 2025-10-13

  articleSenior author
  PublisherDOI

• [Arbuscular mycorrhizal fungi improve physiological metabolism and ameliorate root damage of <i>Coleus scutellarioides</i> under cadmium stress].

  PubMed · 2025-02-25

  article

  .

  PublisherDOI

• An improved end-to-end micro-expression recognition system for real-world videos via dual-input CNN architecture

  Expert Systems with Applications · 2025-05-10

  article
  PublisherDOI

Recent grants

• CAREER: Signal Processing Approaches for Multimedia Security and Information Protection

  NSF · $356k · 2002–2008

• Forensic Hash for Assured Cyber-based Sensing and Communications

  NSF · $344k · 2010–2014

• Addressing Physical-Layer Challenges via CLAWS: Cross-Layer Approaches to Wireless Secure Communications

  NSF · $312k · 2008–2012

• I-Corps Team Proposal "Mini Signal"

  NSF · $50k · 2018–2019

• Collaborative Research: RAPID: Understanding and Facilitating Remote Triage and Rehabilitation During Pandemics via Visual Based Patient Physiologic Sensing

  NSF · $81k · 2020–2022

Frequent coauthors

• Sergios Theodoridis

  National and Kapodistrian University of Athens

  522 shared

• Ahmed H. Tewfik

  Apple (United Kingdom)

  522 shared

• Ali H. Sayed

  École Polytechnique Fédérale de Lausanne

  476 shared

• Walter Kellermann

  Friedrich-Alexander-Universität Erlangen-Nürnberg

  476 shared

• Béatrice Pesquet‐Popescu

  University of Maryland, Baltimore County

  472 shared

• Antonio Ortega

  University of Southern California

  472 shared

• Robert Karam

  University of South Florida

  472 shared

• Athina P. Petropulu

  Rutgers, The State University of New Jersey

  426 shared

Labs

• Communication DepartmentPI

Education

• Ph.D., Electrical Engineering

  Princeton University

  2001

• B.A. Economics, School of Economics & Management

  Tsinghua University

  1996

• B.S.E., Department of Automation

  Tsinghua University

  1996