June 2025 - Gradstudents - mailman.cs.vt.edu

Recruiting participants for research study
by More, Anirudh 20 Jun '25

20 Jun '25

Hello, We are recruiting participants for a research study and would greatly appreciate it if you could distribute the following recruitment message to your grad students who are on campus over the summer: The Systems Cognition Engineering Laboratory at Virginia Tech is conducting a study on dynamic leadership assignment between a human and an autonomous agent in object detection and reassignment tasks employing four autonomous UAVs. We anticipate that the experiment will take roughly 2.5 to 3 hours. Participation is entirely voluntary, and it is not mandatory for students to take part. Participants are compensated for their time at approximately $10 per hour, with up to $10 in bonus. Eligibility Criteria: * Must be a current Virginia Tech student * Have normal or corrected to normal vision. If interested, please contact Anirudh (anirudh1(a)vt.edu). This study has been reviewed and approved by the Virginia Tech Institutional Review Board (IRB #25-350). Thank you for your time and support. Best regards, Anirudh

1 0

Srinivasan Ramachandra Sharma's MS Defense on Monday 06/23 at 11.00 AM in GP 4001
by Sundararajah, Kirshanthan 20 Jun '25

20 Jun '25

All are invited to the MS Thesis Defense of Srinivasan Ramachandra Sharma on Monday 06/23 at 11.00 AM in 4001 Gilbert Place and in my Zoom room https://virginiatech.zoom.us/my/kirshanthans Feel free to invite others as well. Title: Control Flow Merging: A Compiler Transformation to Mitigate Branch Misprediction by Branch Elimination Abstract: Modern superscalar processors have a high theoretical throughput, being capable of executing multiple instructions per clock cycle. In practice, however, this peak is rarely reached because data, structural, and control flow hazards can stall the pipeline and interrupt execution. Branch prediction is widely adopted in modern architectures to mitigate control flow hazards, yet even sophisticated predictors struggle to handle branches driven by random data. Each misprediction flushes speculative instructions and restarts execution on the correct path, making branch mispredictions a major performance bottleneck. Existing compiler solutions, such as outcome precomputation and simple predication, are effective only for loop-invariant conditions or small side-effect-free branches. This work introduces Control Flow Merging (CFM), a compiler transformation that eliminates branches by converting control dependencies into data dependencies. CFM systematically replaces if and if-then-else regions with semantically equivalent, predicated instruction sequences, even when handling operations with side effects. Our work targets branch elimination at the compiler level by converting control flow dependencies into dataflow. We introduce the Control Flow Merging (CFM) pass for branch removal which replaces if and if-then-else regions with predicated instructions, while maintaining semantics and safety, even in the presence of instructions with side effects and implement it in LLVM 14. We also evaluate its impact on real-world benchmarks. Our experiments show that CFM can reduce branch mispredictions by up to 99 % and improve performance. Across the entire benchmark set, the pass delivers upto 53x peedup in the best case, and a geometric-mean speedup of 1.57×. Kirshanthan "Krish" Sundararajah, Ph.D. Assistant Professor Virginia Tech | Computer Science 220 Gilbert St, Suite 4103 Blacksburg, VA 24060 (540) 231-1779 | kirshanthans(a)vt.edu<mailto:kirshanthans@vt.edu> kirshanthans.github.io<https://kirshanthans.github.io/>

1 0

Recruiting for a Research Study on Using Gestures for Learning
by Bruen, Jacqueline 15 Jun '25

15 Jun '25

Hi there, I hope everyone is having a good summer! Please check out my study (happening in Blacksburg this summer and fall) below~ The Mind Music Machine Lab at Virginia Tech is seeking individuals to participate in a study to determine how gestures can improve the acquisition of algorithmic interpretation skills. During this study, participants will: · Complete a demographic survey. · Learn an algorithmic language either with or without accompanying hand gestures over the course of three 30 minute sessions. · Take two tests in a fourth 30 minute session which assesses your acquisition of various algorithmic interpretation skills. · Take a short 10 minute semi-structured interview. You’re eligible for our study if you: · Are over 18 years old · Have normal hearing, vision (or correct-to-normal vision), and fine motor skills The study will take a total of approximately two hours over the course of four 30 minute sessions. Each participant will receive compensation of $10/hour. Total compensation for time and effort for the whole study will be $20 in cash, distributed to the participant at the end of the fourth session. You are free to discontinue participation at any point if you feel uneasy or uncomfortable. Interested individuals can reach out to Jacqueline Bruen at jacquelineb(a)vt.edu. The research protocol for this experiment has been reviewed by the Virginia Tech Institutional Review Board (IRB #24-942). Thank you for your help. Kindly, Jacqueline Elise Bruen Ph.D. Candidate | Mind Music Machine Lab Department of Computer Science Virginia Tech "Mind is the Master power that moulds and makes, And Man is Mind, and evermore he takes The tool of Thought, and, shaping what he wills, Brings forth a thousand joys or a thousand ills." -James Allen

1 0

EduTube AI User Study
by Farzanehpour, Sahar 11 Jun '25

11 Jun '25

Hey Heyyy, 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 and quiz generating might take a bit to provide you the result! 4. Provide your feedback on the video recommendations and the generated quiz. 5. The study will take about 10-15 minutes to complete. 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 Email: saharfarza(a)vt.edu LinkedIn: https://www.linkedin.com/in/saharfarzanehpour/

1 0

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