About

Sudip Vhaduri is an assistant professor at Purdue University in the School of Applied and Creative Computing, formerly known as the Department of Computer and Information Technology. He earned his Ph.D. in Computer Science and Engineering from the University of Notre Dame under the supervision of Prof. Christian Poellabauer. Prior to that, he completed his M.Sc. in Computer Science from the University of Memphis and his B.Sc. in Computer Science and Engineering from the Bangladesh University of Engineering and Technology. His academic lineage traces back through notable scholars including Edsger Wybe Dijkstra. Professor Vhaduri leads the Mobile Artificial Intelligence Laboratory (mAI-Lab), where his research centers on artificial intelligence techniques such as machine learning, deep learning, data mining, and data analytics, applied to mobile and wearable computing. His goal is to develop impactful, sustainable, and secure solutions that empower individuals to improve their quality of life using their own mobile devices interconnected via the Internet-of-Things (IoT). His research integrates three main application areas: health informatics leveraging smartphone and wearable sensing; continuous and implicit user authentication using physiological and behavioral biometrics measured by IoT wearables; and reliable discovery of places of interest using alternative sensor data when location data is unavailable or inaccurate. His work involves analyzing large-scale human study datasets collected through mobile crowdsensing and has been recognized in prestigious outlets such as Forbes Magazine. He leads interdisciplinary collaborative research teams that include experts from sociology, psychology, medical sciences, laryngology, otolaryngology, speech-language pathology, communication sciences and disorders, health sciences, behavioral sciences, statistics, mathematics, and electrical engineering, spanning top academic and industrial research institutions in the US and Europe, including IBM Research. His impactful interdisciplinary and interinstitutional research contributions have earned him consistent top rankings among assistant professor peers in his department at Purdue University and recognition in ScholarGPS. He is a member of several IEEE societies and affiliated with multiple research centers at Purdue, including the Institute for Physical Artificial Intelligence, Applied AI Research Center, CERIAS, PI4D, and the Institute for a Sustainable Future, contributing to the United Nations Sustainable Development Goals.

Research topics

• Computer Science
• Artificial Intelligence
• Computer Security
• Machine Learning
• Medicine
• Psychology
• Internet privacy
• Embedded system
• Internal medicine
• Applied psychology
• Social psychology
• Nursing
• Speech recognition

Selected publications

• Understanding Stability of Choices: Toward Robust Choice-Based Authentication in Cybersecurity

  2026-02-20

  article

  Traditional security questions (e.g., mother's maiden name or first school) are easy to implement and use, but vulnerable to adversarial guessing or research, due to the increasing availability of personal information online, which reduces their effectiveness in cybersecurity authentication. Choice-based security questions offer a promising alternative by leveraging stable personal preferences that are harder for adversaries to predict. Prior studies have not comprehensively examined the temporal stability and consistency of users' responses to a wide range of choice-based questions and their potential to improve the authentication. In this work, we study and analyze responses from 276 participants to nine choice-based questions (e.g., favorite color, number, season, day, animal, sport, month, food, and artist), collected three times with one-month intervals. Results show that questions on color, number, season, day, and animal exhibit high stability over a 2 -month period, while maintaining high uncertainty against adversarial guessing, making them reliable candidates for robust authentication. In contrast, sport, month, food, and artist questions demonstrate lower stability. These findings highlight the importance of balancing stability (in terms of user memorability and recall consistency) with uncertainty (in terms of resistance in adversarial prediction) when designing choice-based questions, enabling more secure, user-friendly authentication mechanisms in cybersecurity systems.

  PublisherDOI

• Investigating the Role of the Instructional Team in Enhancing Student Teamwork Experiences in an Introductory Cybersecurity Course

  2025-11-02

  article

  This innovative practice full research paper explores the role of the instructional team in supporting teamwork within a large, project-based course for first-year cybersecurity students. The study investigates how instructional teams can enhance collaboration and problem-solving skills critical for student success. Guided by two research questions, the study examines: (1) How do students perceive the effectiveness of instructional team support in facilitating teamwork? and (2) What aspects of instructional support shape students' collaboration experiences? At semester's end, 141 students completed a survey reflecting on their teamwork experiences and the instructional team's role. Thematic analysis of the qualitative responses revealed that students highly valued clear guidelines and expectations provided by the instructional team, which improved assignment understanding and coordination. Constructive feedback helped students refine their work, particularly on team milestones. Responsiveness and timely communication were frequently praised, with students noting the importance of quick, helpful replies. The instructional team also assisted with conflict resolution and interpersonal challenges, ensuring a productive team environment. Office hours were especially important for interpreting feedback and improving outcomes. Additional support included teaching teamwork strategies and offering emotional encouragement, which students found motivating. However, students suggested improvements such as clearer, more actionable feedback, refined rubrics, additional templates or examples, and more proactive outreach. These findings underscore the critical influence of a responsive instructional team in fostering collaboration in projectbased learning. By addressing students' feedback, instructional teams can more effectively support the development of essential teamwork and professional skills in STEM education.

  PublisherDOI

• Mapping Pilot Stress and Fatigue During Flight Sessions

  Purdue e-Pubs (Purdue University System) · 2025-01-01

  article

  This study examined physiological and self-reported indicators of stress and fatigue in four student pilots during training flights at Purdue University. Each pilot wore a wrist sensor that recorded heart rate, electrodermal activity, temperature, and acceleration, and completed pre and post-flight ratings of stress, fatigue, and sleep questionnaire. The pilots flew different types of training sessions, providing variation in environmental and flight conditions. All physiological data were cleaned and standardized before analysis. Across pilots, temperature and electrodermal activity generally increased during flight, showing a gradual rise in arousal that can come from several factors, including heat, physical effort, and normal physiological activation. Heart rate showed shorter, sharper changes that tended to appear during more intense moments in the flight, suggesting it may be more sensitive to brief shifts in stress. Self-reported fatigue increased for all participants, showing a pattern similar to the gradual rise seen in temperature and electrodermal activity, while stress ratings remained mostly steady. Sleep questionnaire results indicated generally healthy long-term sleep habits, lowering the chance of major confounding from chronic sleep issues. Overall, the physiological and subjective measures together suggest that slower trends in the signals may relate to developing fatigue, while quicker heart-rate changes may reflect momentary stress, though these interpretations are not definitive given the small sample size.

  Publisher

• A Comprehensive Survey of Challenges and Opportunities of Few-Shot Learning Across Multiple Domains

  ArXiv.org · 2025-04-05

  preprintOpen accessSenior author

  In a world where new domains are constantly discovered and machine learning (ML) is applied to automate new tasks every day, challenges arise with the number of samples available to train ML models. While the traditional ML training relies heavily on data volume, finding a large dataset with a lot of usable samples is not always easy, and often the process takes time. For instance, when a new human transmissible disease such as COVID-19 breaks out and there is an immediate surge for rapid diagnosis, followed by rapid isolation of infected individuals from healthy ones to contain the spread, there is an immediate need to create tools/automation using machine learning models. At the early stage of an outbreak, it is not only difficult to obtain a lot of samples, but also difficult to understand the details about the disease, to process the data needed to train a traditional ML model. A solution for this can be a few-shot learning approach. This paper presents challenges and opportunities of few-shot approaches that vary across major domains, i.e., audio, image, text, and their combinations, with their strengths and weaknesses. This detailed understanding can help to adopt appropriate approaches applicable to different domains and applications.

  PublisherOA PDFDOI

• Challenges and Opportunities of Federated Learning In the Age of IoT: A Multi-Domain Comprehensive Survey

  2025-09-04

  preprintOpen accessSenior author

  The increase of machine learning (ML) across various industries and the usage of Internet of Things (IoT) devices in business operations creates a need for data to be constantly fed into ML models to upgrade their mechanisms and accuracy. Because of these models' need for an extensive amount of data, many organizations tend to gather information from users, raising concerns about data security and privacy for individuals whose data may be collected and stored without their knowledge. Individuals should have control over their data and the right to hold the raw data on their side while getting updated models from the server that organizations use. Unlike traditional ML modeling, which centralizes sensitive data into a single server, federated learning (FL) models distribute the training process across multiple devices/nodes. FL models, therefore, offer the greatest assistance with data security and privacy. This paper provides a comprehensive understanding of different FL approaches that are applicable to three major domains, i.e., audio, image, and text, and their combinations, along with their challenges and opportunities, and different IoT applications to guide future researchers.

  PublisherOA PDFDOI

• Biometrics for Wearable Devices

  2025-01-01

  book-chapter1st authorCorresponding
  PublisherDOI

• Multimodal Audio-Based Disease Prediction With Transformer-Based Hierarchical Fusion Network

  IEEE Transactions on Audio Speech and Language Processing · 2025-01-01 · 2 citations

  article

  Audio-based disease prediction is emerging as a promising supplement to traditional medical diagnosis methods, facilitating early, convenient, and non-invasive disease detection and prevention. Multimodal fusion, which integrates features from various domains within or across bio-acoustic modalities, has proven effective in enhancing diagnostic performance. However, most existing methods in the field employ unilateral fusion strategies that focus solely on either intra-modal or inter-modal fusion. This approach limits the full exploitation of the complementary nature of diverse acoustic feature domains and bio-acoustic modalities. Additionally, the inadequate and isolated exploration of latent dependencies within modality-specific and modality-shared spaces curtails their capacity to manage the inherent heterogeneity in multimodal data. To fill these gaps, we propose a transformer-based hierarchical fusion network designed for general multimodal audio-based disease prediction. Specifically, we seamlessly integrate intra-modal and inter-modal fusion in a hierarchical manner and proficiently encode the necessary intra-modal and inter-modal complementary correlations, respectively. Comprehensive experiments demonstrate that our model achieves state-of-the-art performance in predicting three diseases: COVID-19, Parkinson's disease, and pathological dysarthria, showcasing its promising potential in a broad context of audio-based disease prediction tasks. Additionally, extensive ablation studies and qualitative analyses highlight the significant benefits of each main component within our model.

  PublisherDOI

• Enhancing Teamwork in Project-Based Learning: Challenges, Reflections, and Strategies for Improvement

  2025-11-02

  article

  This Innovative Practice, Full Paper focuses on the teamwork experiences of first-year students in a large, projectbased learning course on Cybersecurity. The motivation behind this research is to understand the challenges students face while working in teams and to identify strategies for improving team dynamics, which are essential for academic and professional success. Guided by two research questions-(1) What challenges did students face while working in teams in an introductory Cybersecurity course? and (2) What strategies did students propose to improve team dynamics?-this study collected survey responses from 141 first-year students enrolled in the course.Thematic analysis of the qualitative responses revealed key challenges: scheduling conflicts, communication issues, uneven contribution, procrastination, and lack of accountability. Students suggested several strategies to improve team functioning, such as clear communication protocols, early role definition, regular team meetings, and structured time management. They also emphasized the importance of building interpersonal trust through informal bonding and fostering shared responsibility through peer feedback and conflict resolution mechanisms. Findings from this study indicate that team-based learning can be enhanced through proactive instructional support. This includes setting clear expectations, offering structured check-ins, and integrating peer evaluation tools. By addressing these common barriers, educators can foster more effective and equitable teamwork experiences. The study contributes practical recommendations for improving collaborative learning and calls for future research to examine the implementation and impact of these strategies across varied educational contexts

  PublisherDOI

• Detecting Patient and Healthy People’s Personalized Breathing Patterns with Few-Shot Learning

  2025-12-03

  article

  Analysis of respiratory sounds, such as coughing and breathing, has emerged as a promising non-invasive approach for the early detection of pulmonary conditions, including COVID-19 and chronic obstructive pulmonary disease (COPD), as well as for managing treatment plans. In this work, we explore audio-based classification of respiratory conditions in terms of breathing patterns using the few-shot learning approach that requires only a few samples to develop models. We experimented with three publicly available datasets of audio recordings of breathing patterns of healthy people and patients with COVID-19 or COPD. Through a detailed evaluation using three types of audio features commonly employed for audio event classification, with varying embedding dimensions and shot numbers, we found that Mel-Frequency Cepstral Coefficients (MFCCs) with 20 embedding dimensions can achieve an average accuracy of around 85% using only 10 shots when classifying the breathing patterns obtained from the three datasets. These findings highlight the potential for developing audio-based screening tools that require only a few samples, which can be utilized for public health diagnostics.

  PublisherDOI

• Challenges and Opportunities of Generative Ai Models in Audio/Acoustic

  SSRN Electronic Journal · 2025-01-01

  preprintOpen accessSenior author
  PublisherDOI

Frequent coauthors

• Aaron Striegel

  University of Notre Dame

  26 shared

• Mingyue Ji

  Sapienza University of Rome

  25 shared

• Olga Saukh

  25 shared

• Alessandro Mei

  25 shared

• Mingjun Xiao

  University of Science and Technology of China

  25 shared

• Shaowei Wang

  25 shared

• Yun Lin

  Rutgers Sexual and Reproductive Health and Rights

  25 shared

• Dimitrios Koutsonikolas

  Universidad del Noreste

  25 shared

Awards & honors

• Best Paper Runner-Up Award at IEEE/ACM CHASE 2022