- Gradstudents - mailman.cs.vt.edu

Invitation to Research Defense of Satvik Chekuri on 6/18 at 3:30pm in 2030 Torgersen Hall and in my Zoom room; pilot testing!
by Fox, Edward 11 Jun '25

11 Jun '25

All are invited to the Research Defense of Satvik Chekuri on 6/18 at 3:30pm in 2030 Torgersen Hall and in my Zoom room https://virginiatech.zoom.us/my/ed.fox Feel free to invite others as well. PS Pilot testing for the user study is commencing, so let Satvik know if interested. -- Edward A. Fox, Ph.D., Emeritus Professor of Computer Science Virginia Tech | CS (0106) 1160 Torgersen Hall (CS dept. office) Blacksburg, VA 24061 (540) 231-5113 | fox(a)vt.edu<mailto:fox@vt.edu> | http://fox.cs.vt.edu<http://fox.cs.vt.edu/> = = = see title and abstract below = = = Title: A Scholarly Information System for Long Documents and their Elements - Status Report for Research Defense Abstract: Scholarly information systems for long documents, such as electronic theses and dissertations (ETDs), offer significant potential to assist researchers, students, and domain experts in navigating complex scholarly content. In this work, we present a scalable Information Retrieval system for ETDs and their elements (ETD-IR System), designed to facilitate enriched scholarly research by supporting both metadata-level and content-level exploration across ETD collections. Building upon this foundation, we further develop a Knowledge Graph (KG)-augmented Question Answering (QA) system designed specifically for ETDs, integrating multiple innovations across knowledge representation, retrieval, and answer generation. We construct an ETD Knowledge Graph that captures both high-level metadata and fine-grained content structure, including classified abstract sentences (Background, Objective, Method, Result, Conclusion), figure captions, table captions, and topics, derived through a combination of automated extraction, topic modeling, and abstract sentence classification. To support realistic QA scenarios, domain-specific questions are automatically generated by leveraging expert bios, enabling contextually relevant question sets for evaluation.

1 0

Fw: Tenure-Track Faculty Openings
by Lee, Sang Won 10 Jun '25

10 Jun '25

Forwarding as requested. Sang [cid:f6565c55-e106-4778-825c-a197453bcb71]<https://outlook.office.com/bookwithme/user/4cafba6be9af4b5081ea587bbb694b20…> Book time to meet with me<https://outlook.office.com/bookwithme/user/4cafba6be9af4b5081ea587bbb694b20…> ________________________________ From: jiang.xu(a)hkust-gz.edu.cn <jiang.xu(a)hkust-gz.edu.cn> Sent: Tuesday, June 10, 2025 7:07 PM To: Lee, Sang Won <sangwonlee(a)vt.edu> Subject: Tenure-Track Faculty Openings Dear Prof. Lee, Hope this message finds you a good summer! We are hiring tenure-track faculty in device and fabrication, circuit, architecture and system, and EDA areas with highly competitive salaries in the world and cutting-edge research facilities. I'd appreciate you recommending your students to us. Information on our current faculty and the openings are in the following links. https://facultyprofiles.hkust-gz.edu.cn/thrust-faculties?code=10011A1000000… https://personal.hkust-gz.edu.cn/jiangxu/tenure-track%20openings.pdf Best, Prof. Jiang XU ----------------------------------- Department Head of Microelectronics Thrust Hong Kong University of Science and Technology (Guangzhou) personal.hkust-gz.edu.cn/jiangxu <https://personal.hkust-gz.edu.cn/jiangxu>

1 0

EduTube AI User Study
by Farzanehpour, Sahar 02 Jun '25

02 Jun '25

Hey everyone, We’re conducting a study to explore how learners use and benefit from EduTube AI, a tool that recommends educational videos and generates a quiz to help you find related videos and learn algorithm topics more effectively. Survey Link: https://virginiatech.questionpro.com/edutube If you choose to participate, here’s what we’ll ask you to do: 1. Visit the EduTube AI tool link (https://edutube.cs.vt.edu/). 2. Choose a topic you’d like to learn (e.g., Dijkstra’s Algorithm) and briefly describe it along with your current knowledge level. 3. The search might take a bit to provide you the result! 4. Provide your feedback on the video recommendations and the generated quiz. Participation is voluntary and open to individuals 18 years or older. You can withdraw at any time. All responses will be kept confidential and reported in aggregate, without identifying individuals. This study has been approved by IRB with 25-480 number. If you have any questions, feel free to contact: * Sahar Farzanehpour – saharfarza(a)vt.edu<mailto:saharfarza@vt.edu> * Dr. Chris Brown – dcbrown(a)vt.edu<mailto:dcbrown@vt.edu> Thank you for your time and support. We truly appreciate your help! Best regards, -- Sahar Farzanehpour Graduate Student, Computer Science Department Virginia Tech LinkedIn: https://www.linkedin.com/in/saharfarzanehpour/

1 0

Akshay Bansal's PhD defense, Tuesday, May 27, 2025, 9am
by Sikora, Jamie 23 May '25

23 May '25

Happy Friday! I welcome everyone to attend Akshay Bansal’s PhD defense next week. Details below. Have a great long weekend! -Jamie > Title: Adversarial two-party quantum interactions in cryptography and machine learning > Date/Time: 05/27/2025, 9:00 AM > Location: 3001 Gilbert Place. Zoom option available upon request.

1 0

IMPORTANT* Flight Related Travel Changes
by Teresa Hall 22 May '25

22 May '25

PLEASE READ *BOTH* UPDATES BELOW! *Travelers with a base point of Blacksburg, VA or Roanoke, VA* may elect to fly from any airport within 300 miles without a documented cost comparison. Airports within 300 miles include but are not limited to: * Charlotte International Airport * Dulles International Airport * Piedmont Triad International Airport (Greensboro) * Raleigh-Durham International Airport * Richmond International Airport * Ronald Regan Washington National Airport *Premium economy and business-class fares * * Economy-class tickets must be reserved for all flights with an in-air travel time of less than eight (8) consecutive hours. * Premium economy-class tickets may be reserved for flights with an in-air travel time that exceeds eight (8) consecutive hours. * Business-class tickets may be reserved for long-haul flights with an in-air travel time that exceeds ten (10) consecutive hours when there is also a business meeting conducted within three hours of landing. **Note: The term “consecutive hours” applies to one flight from take-off to landing. The traveler may not combine in-air travel times on a multi-leg itinerary to meet the requirements for the purchase of premium economy or business-class tickets. *Teresa Hall** **Fiscal Technician* Virginia Tech| Department of Computer Science Torgersen Hall, 1180E (0106) 620 Drillfield Drive Blacksburg, VA 24061 540-231-8454 | teresa(a)vt.edu <mailto:teresa@vt.edu> Office Hours: Monday, Wednesday, &Thursday Telework:Tuesday &Friday __

1 0

Fw: Xrai Lab Call for Participants – VR Teams in Military Battle Drill
by Coulson, Sara 19 May '25

19 May '25

Sharing as requested... Sara Coulson Director of Graduate Programs | Computer Science Torgersen 3210C 620 Drillfield Drive Blacksburg, VA 24061 ________________________________ From: McMahan, Ryan <rpm(a)vt.edu> Sent: Monday, May 19, 2025 4:08 PM To: Coulson, Sara <sara83(a)vt.edu> Subject: Xrai Lab Call for Participants – VR Teams in Military Battle Drill Hi Sara, Can you share the IRB-approved Call for Participants below and attached flyer with the graduate CS students? Best regards, Ryan Ryan P. McMahan, Ph.D. Deputy Director, Center for Human-Computer Interaction (CHCI) Professor, Department of Computer Science (CS) Virginia Tech CALL FOR VR TEAMS! Experience a Military Battle Drill! The Extended Reality & Artificial Intelligence (Xrai) Lab at Virginia Tech is inviting you to participate in a research study focused on figuring out how groups of people and virtual humans perform as a military team in virtual reality (VR). The data collected will be used to develop multimodal and machine learning models of team performance based on a VR battle drill simulation. Eligible participants: * Are 18 to 64 years old, * Have 20/20 vision without or with contacts (glasses are not permitted during the study), * Can walk and use both hands, * Are willing to attend a 2-hour in-person experimental session in Gilbert Place, * Are willing to participate as a team with up to three other participants, * Are proficient in reading, writing, listening, and speaking English, * Are U.S. Citizens, * Are NOT pregnant, * Do NOT have pre-existing binocular vision abnormalities or psychiatric disorders, * Have NOT recently undergone any medical procedure (including cosmetic procedures), and * Do NOT suffer from a heart condition or other serious medical condition. Participants will receive a $45 Amazon gift card in compensation for participating in and completing the study, if they are not military personnel. If you are interested in more information, please review the consent form document for the study and verify your eligibility here: https://forms.gle/mFMMrzLx2NuztWZa8. After completing the online survey, the research team will reach out to you via email to confirm a time for the research session. This project is supervised by Dr. Ryan P. McMahan (rpm(a)vt.edu<mailto:rpm@vt.edu>, 540-231-9722) from the Department of Computer Science. This experiment has been approved, as required, by the Virginia Tech Institutional Review Board.

1 0

Fw: Postdoc Opportunity - ARL Distinguished Postdoctoral Fellowships - Please Share!
by Cho, Jin-Hee 13 May '25

13 May '25

Dear PhD students and faculty colleagues, I wanted to share information about an excellent postdoctoral fellowship program—please see the details below. Faculty, feel free to forward this to anyone who may be interested. Best regards, Jin-Hee ________________________________ From: NASEM Fellowships Office Outreach Sent: Tuesday, May 13, 2025 9:05 AM To: Cho, Jin-Hee Subject: Postdoc Opportunity - ARL Distinguished Postdoctoral Fellowships - Please Share! You don't often get email from fellowshipsofficeoutreach(a)nas.edu. Learn why this is important<https://aka.ms/LearnAboutSenderIdentification> [https://outlook.office.com/mail/inbox/id/AAkALgAAAAAAHYQDEapmEc2byACqAC%2FE…] Dear Sir or Madam: We ask your assistance in forwarding this message to graduate students in your department to inform them of an outstanding fellowship opportunity with the Army Research Laboratory (ARL). The Army Research Laboratory (ARL) invites exceptional early career researchers to apply for an ARL Distinguished Postdoctoral Fellowship<http://pull.xmr3.com/p/6373-D47D/202680950/1dd47577-2f43-49b6-a582-852b9019…> that provides recipients the opportunity to pursue independent research of their own choosing that supports the mission of ARL<http://pull.xmr3.com/p/6373-177C/202680954/5cf78680-1db1-4747-b228-4cc6e360…>. Candidates must display extraordinary ability in scientific research, show clear promise of becoming outstanding leaders, and are expected to have already successfully tackled a major scientific or engineering problem during their thesis work or to have provided a new approach or insight, evidenced by a recognized impact in their field. Eligibility and Selection Criteria * Must have completed all the requirements for a Ph.D. or Sc.D. degree in the physical sciences, life sciences, computational sciences, behavioral sciences, or engineering by October 1, 2025. Applicants who intend to defend their dissertation after October 1, 2025, are not eligible. * May not be more than five years beyond the award date of their Ph.D. or Sc.D. by the fellowship application deadline (July 10, 2025 at 5:00 PM Eastern Time).\ * Must demonstrate exceptional qualifications with respect to academic and scholarly achievement, as evidenced by research and publication. * Are expected to have conducted research on a major scientific or engineering problem during their thesis work or have provided a new approach or insight, evidenced by a recognized impact in their field. There are no citizenship requirements; however, selected fellows must pass a Department of Defense (DoD) clearance process, requiring a background security investigation. Foreign nationals must get further DoD approval, requiring a background investigation. Fellowship benefits * Full-time, one-year in-residence appointment at ARL, renewable for up to three years based on performance * Annual stipend of $115,000 * Health insurance (including dental and vision) * Paid relocation and professional travel allowance Interested candidates can find additional information and apply at https://arldpf.nationalacademies.org<http://pull.xmr3.com/p/6373-DA17/201262660/1dc62db9-c1a7-4621-adc6-4fd2dd3d…>. Sincerely yours, Elizabeth M. Prescott, D.Phil. Director, Fellowships Office [https://outlook.office.com/mail/inbox/id/AAkALgAAAAAAHYQDEapmEc2byACqAC%2FE…] ________________________________ If you prefer not to receive additional messages from this sender, please Click here<http://www.xmr3.com/rm/1458401-63182901-2-6373-US3-C74B/JICHO@VT.EDU/HCS87FE> and confirm your request.

1 0

Andrew Zhang's MS defense, 12:00 PM, May 13th
by Thomas, Chris 13 May '25

13 May '25

Hello, You are kindly invited to attend the MS thesis defense of Andrew Zhang, May 13th, 12:00 PM. Details follow. Title: Discrete Diffusion for Text Infilling Date: 05/13/25 Time & Location: 12:00 pm, Virtual Zoom: https://virginiatech.zoom.us/j/8956885181 Committee Members: Chris Thomas, Pinar Yanardag, Xuan Wang Abstract: Discrete diffusion models are a new class of text generators which offer advantages such as bidirectional context use, parallelizable generation, and flexible prompting compared to autoregressive models. However, existing text diffusion models are unable to handle flexible-length text infilling tasks without ground-truth positional information. To address this limitation, we propose Discrete Diffusion with Optimal Transport Position Coupling (DDOT), the first discrete diffusion model capable of flexible-length text infilling. DDOT introduces a diffusion process that simultaneously denoises discrete token values and continuous token positions with a sample-level optimal transport (OT) coupling to ensure smooth token flows throughout the diffusion process. Our method is orthogonal to existing discrete text diffusion methods and is compatible with various pretrained text denoisers. Extensive experiments on constrained text generation benchmarks demonstrate that DDOT achieves competitive performance among non-autoregressive models, approaches autoregressive methods, and leads to significant improvements in training efficiency and flexibility. Thank you! Chris Thomas

1 0

Anushka Sivakumar's MS defense, 10:30 AM, May 13th
by Thomas, Chris 13 May '25

13 May '25

Hello, You are kindly invited to attend the MS thesis defense of Anushka Sivakumar, May 13th, 10:30 AM. Details follow. Title: Model Control through Lightweight Activation Steering for Vision Language Models Date: 13th May 2025 Time & Location: 10:30am-12:00pm, Data and Decision Sciences (Zoom) Zoom: https://virginiatech.zoom.us/j/8956885181 Committee Members: Dr. Chris Thomas, Dr. Pinar Yanardag, Dr. Xuan Wang Abstract: This work proposes a lightweight steerer, a parameterized model that guides large vision language models (VLMs), toward outputs that better adhere to desired instructions. Our aim is to achieve effective model control, steering behavior towards a target and away from its converse, using paired prompts. We introduce a robust, lightweight steering module that operates in the activation space, adapting steering to VLMs by connecting the language modality with the image context. This approach enables inference-time control over complex output semantics, such as entailment, hallucination, tone, and perspective, and does not directly modify model weights thereby preserving performance on off-target tasks. A key feature of our method is a layer-agnostic steering approach that does not require parameter tuning to identify optimal steering layers or pre-extracted steering vectors to capture complex relationships between target and converse prompts. In addition to our proposed steering module, we introduce a novel multimodal steering dataset, VNIA (Visual Narrative Intent Alignment), designed to facilitate the development and evaluation of vision-language model steering. Thank you! Chris Thomas

1 0

Swapnil Singh's MS defense, 1:00 pm Wednesday 5/14
by Zhang, Liqing 12 May '25

12 May '25

Dear Colleagues and Students, I am delighted to invite you to Swapnil Singh's MS defense on Wednesday, May 14. Title: Multi-modal Multi-Level Neuroimaging Fusion with Modality-Aware Mask-Guided Attention and Deep Canonical Correlation Analysis to Improve Dementia Risk Prediction Date: Wednesday May 14, 2025 Time: 1:00 pm EST Location: Torgersen Hall, Room 1210 Zoom link: https://virginiatech.zoom.us/j/81949848609 Committee Members: Liqing Zhang (co-Chair), Da Ma (co-Chair), T M Murali, Chris Thomas Abstract: Alzheimer’s Disease (AD) is a progressive neurodegenerative disorder characterized by structural and molecular changes in the brain. Early diagnosis and accurate subtyping are essential for timely intervention and therapeutic planning. This thesis presents a novel multimodal deep learning framework that integrates T1-weighted MRI and Amyloid PET imaging to improve the diagnosis and stratification of AD. The proposed architecture leverages a two-stage pipeline involving modality-specific feature extraction using ResNet50 backbones, followed by middle fusion enhanced with a Modality-Aware Mask-Guided Attention (MAMGA) mechanism. To address missing modalities and inter-modal misalignment, the model incorporates Random Modality Masking and Deep Canonical Correlation Analysis (DCCA) for cross-modal feature alignment. Experiments on the ADNI dataset demonstrate that the proposed MRI+PET (MAMGA+DCCA) model achieves a balanced accuracy of 0.998 and an AUC-ROC of 0.999 in distinguishing stable normal cognition (sNC) from stable Alzheimer's Disease (sDAT). For the more challenging task of separating stable and progressive MCI (sMCI vs. pMCI), the best-performing fusion model achieved a balanced accuracy of 0.732 and an AUC of 0.789. Extensive ablation studies confirm the contributions of MAMGA, DCCA, and dual-optimizer strategies in enhancing diagnostic robustness. This work highlights the clinical potential of multimodal deep learning frameworks in improving early Alzheimer's detection and stratification. -- Liqing Zhang, Ph.D. Professor, Department of Computer Science, Virginia Tech Fralin Biomedical Research Institute, Virginia Tech FBRI Cancer Research Center, Washington, DC Professor of Health Sciences, Virginia Tech

1 0