June 2023 - Gradstudents - mailman.cs.vt.edu

Jie Bu's Ph.D. defense [Feb 9 at 4 pm]
by Anuj Karpatne 02 Dec '24

02 Dec '24

Hello everyone, You are cordially invited to the final defense of PhD student Jie Bu on Thursday, Feb 9 at 4 pm. Zoom Link: https://virginiatech.zoom.us/j/5142223333 Title: Achieving More with Less: Learning Generalizable Neural Networks With Less Labeled Data and Computational Overheads Committee: Dr. Anuj Karpatne (advisor) Dr. Adrian Sandu Dr. Ismini Lourentzou Dr. Naren Ramakrishnan Dr. Anish Arora (from Ohio State University) Abstract: Recent advancements in deep learning have demonstrated its incredible ability to learn generalizable patterns and relationships automatically from data in a number of mainstream applications. However, the generalization power of deep learning methods largely comes at the costs of working with very large datasets and using highly compute-intensive models. Many applications cannot afford these costs needed to ensure generalizability of deep learning models. For instance, obtaining labeled data can be costly in scientific applications, and using large models may not be feasible in resource-constrained environments involving portable devices. This thesis aims to improve efficiency in machine learning by exploring different ways to learn generalizable neural networks that require less labeled data and computational resources. We demonstrate that using physics supervision in scientific problems can reduce the need for labeled data, thereby improving data efficiency without compromising model generalizability. Additionally, we investigate the potential of transfer learning powered by transformers in scientific applications as a promising direction for further improving data efficiency through domain adaptation. On the computational efficiency side, we present two efforts for increasing parameter efficiency of neural networks through novel architectures and structured network pruning. Thanks, -Anuj ------------------- Anuj Karpatne Assistant Professor Computer Science, Virginia Tech https://people.cs.vt.edu/karpatne/

2 6

Room Available in Torgersen Hall for CS Graduate Students
by Mayo, Trey 29 Aug '23

29 Aug '23

A reminder for all that in spring 2022 we opened a room in Torgersen Hall available exclusively for CS Graduate Students. Torgersen Hall 3200 (TORG 3200) is located directly across from the CS Office of Graduate Programs (TORG 3210). Photos are attached. It provides space for students to meet as a group to work on projects or individuals who just need to a quiet place to study or hang out between classes. The room should not be used for personal/group meetings/parties or classes/tutoring sessions. It is not a reservable space. Access is available to any currently enrolled CS graduate student regardless of degree program or campus. Access to the room is controlled with your Hokie Passport (Student ID card). The room is available 24/7 as long as Torgersen Hall is open. If you have trouble accessing the room with your Hokie Passport, please visit the CS Office of Graduate Programs in TORG 3210 and we'll research the matter. (We believe we've resolved all previous access issues reported). Please note that this room is only for CS Graduate Students. If we find an abuse of the facilities, we will investigate and remove access. Please do your part as a responsible Hokie and help us keep the room clean and in good shape. We hope you enjoy. ============================================================== Trey Mayo, Ed.D. Director of Graduate Programs Virginia Tech | Department of Computer Science Torgersen Hall Suite 3210 620 Drillfield Drive (MC 0106) | Blacksburg, VA 24061 (540) 231-0780 | treymayo(a)vt.edu<mailto:treymayo@vt.edu> | http://cs.vt.edu<http://cs.vt.edu/>

1 2

CS Course Changes Beginning in Fall 2023
by Cliff Shaffer 27 Jun '23

27 Jun '23

Hello! I hope that you are having a good summer. This email is to tell you about some course numbering changes that are about to take place. These changes are mostly related to a major reorganization of our offerings in the general area of Artificial Intelligence and Machine Learning. We hope as a result of these changes our course offerings are expanded, the prerequisite structure is more sane, and some prior confusion and duplications in offerings is eliminated. Here is the short version of the actual changes taking place for this Fall: - Three courses currently offered for Fall as CS5914 are changing to their permanent numbers. These are CS5614 Natural Language Processing taught by Dr. Wang; CS5864 Learning-based Computer Vision taught by Dr. Thomas, and CS5784 Software Project Management taught by Dr. Kennedy. (OK, that last is not AI or ML, but it is a newly permanent course.) - The sections of CS5525 Data Analytics I taught by Drs. Jafari and Skarzynski will be replaced by CS5805 Machine Learning I. - CS5824 Advanced Machine learning taught by Dr. Eldardiry will be replaced by CS5806 Machine Learning II. Students in these sections will automatically be moved to the newly numbered sections (in some cases, even the CRNs will stay the same). Most people won’t need to do anything. Here is the fine print involved. - Going forward, CS5805 Machine Learning I will be the prereq for CS5806 Machine Learning II, CS5814 Deep Learning, CS5614 NLP, and CS5864 Learning-based Computer Vision. - If you are currently scheduled for Dr. Eldardiry’s section of CS5824, you should only take the new 5806 if you have had some ML course in the past (senior level is fine). If you have no ML background, then you should take CS5805 Machine Learning I instead, and take CS5806 in the future if you want (or any of the other courses listed above that CS5805 is the prereq for). - CS5525 and CS5526 are going away. STAT might continue to teach both in some form, we don’t particularly recommend that you take STAT/CS5525 from them. - We do not recommend that you take ECE5424 (currently cross-listed as CS5824). Take CS5806 instead. - Until now, CS6524 Deep Learning was cross-listed with an ECE course. CS6524 is going away. We do not recommend that you take the ECE version, but instead take CS5814. - Students may not get credit for both CS5525 and CS5805. - Students may not get credit for both CS5824 and CS5806. Students currently enrolled in one of the sections of CS5824 taught by ECE this Spring will be dis-enrolled from those sections, since they are going away. You can request to be added to CS5806 taught by Dr. Eldardiry instead, you will be given priority for a slot. Or you can ask ECE to let you into the equivalent section of ECE5424. They might or might not have room. Finally, if you have already taken any of these courses (such as the now deprecated CS5525 or CS5824), that is completely fine. They continue to count toward your degree or certificate. So long as you don’t try to get credit for both the old version and the new version of a course, everything should count for you. If for some reason you need to take the ECE or STAT version of a course, we will likely be able to count that on your plan of study. But you will need to check with us about the details. -- Dr. Cliff Shaffer Professor and Associate Department Head for Graduate Studies Department of Computer Science Phone: (540) 231-4354 Virginia Tech, Blacksburg, VA 24061 WWW: www.cs.vt.edu/~shaffer

1 0

ACTORS NEEDED FOR A SIMULATED JOB INTERVIEW RECORDING
by Jelson 22 Jun '23

22 Jun '23

We are recruiting seniors and graduate students who are actively looking for a job to participate as an actor in our research study. We are conducting a research study investigating eye contact in remote job interviews. We need actors to produce recordings of a simulated job interview for our study. All actors will be expected to bring a job posting that they are targeting from a site like LinkedIn. You will then be asked to simulate a job interview for that company. Actors will get a $25 Amazon gift card. The study is being conducted under Dr. Sang Won Lee and EchoLab. We anticipate it will take 1.5 hours to complete the recording. Eligibility - Currently looking for a full-time position - Conducted in-person or Zoom interview recently (within three months) - Native in English - Comfortable participating in a mock interview and recordings being used for research purposes. - (preferred) Acting/public speech background If you are interested in participating, please fill out the following survey here <https://forms.gle/k55EPDUxfC6jP9i3A>.

1 0

Ming Zhu's Ph.D. defense - June 22, 3:30PM
by Ismini Lourentzou 21 Jun '23

21 Jun '23

Hello everyone, You are cordially invited to Ming Zhu's Ph.D. defense *Title*: Neural Sequence Modeling for Domain-Specific Language Processing: A Systematic Approach *Committee*: Dr. Ismini Lourentzou (Co-chair), Dr. Danfeng (Daphne) Yao (Co-chair), Dr. Edward A. Fox, Dr. Chris Brown, Dr. Wasi Uddin Ahmad *Date and Time:* Thursday, June 22, 2023, 3:30PM *Zoom*: https://virginiatech.zoom.us/j/88565467370 <https://www.google.com/url?q=https://virginiatech.zoom.us/j/88565467370&sa=…> *ABSTRACT*: In recent years, machine learning for natural language processing (ML4NLP) has made substantial progress in many tasks, including information retrieval, question answering, information extraction, machine translation, etc. Benefiting from the highly scalable attention-based Transformer architecture and enormous open access online data, large-scale pre-trained language models have shown great modeling and generalization capacity for sequential data. However, not all domains benefit equally from the rapid development of ML4NLP. Domains like healthcare and software engineering have vast amounts of sequential data containing rich knowledge, yet remain under-explored due to a number of challenges: 1) the distribution of the sequences in specific domains is different from the general domain; 2) the effective comprehension of domain-specific data usually relies on domain knowledge; and 3) the labeled data is usually scarce and expensive to get in domain-specific settings. In this thesis, we focus on the research problem of applying ML4NLP methods to address both common and domain-specific challenges from the healthcare and software engineering domains. We systematically investigate neural-based machine learning approaches to address the above challenges in three research directions: 1) learning with long sequences, 2) learning from domain knowledge, and 3) learning under limited supervision. Our work can also potentially benefit more domains with large amounts of sequential data. Best, Ismini -- *Ismini Lourentzou* <https://isminoula.github.io/> Assistant Professor Virginia Tech | Computer Science Email: ilourentzou(a)vt.edu

1 0

Participants needed for a Comparative Study of Body Doubling in Extended Reality (IRB # 23-144)
by Swetha Annavarapu 15 Jun '23

15 Jun '23

Hello everyone, Please see below details for a user study opportunity. Participants are needed for a HCI research study (IRB # 23-144) to compare the effectiveness of Body Doubling in Mixed Reality. Body doubling is a productivity strategy where you work in the presence of another person to increase focus and avoid distractions. As part of the study participants will be asked to perform a task and may be using a mixed reality headset. You are eligible to participate if you are in the age group of 18-60 years, English speaking, not pregnant and able to provide consent. The duration of the study is approximately 60 minutes. Each participant will be given a $5 gift card at the end of the study for their time and effort for participating in the user study. Participants are required to come to Gilbert Place Building, Blacksburg for the user study. *For more information or scheduling a user study, please contact Swetha Annavarapu at aswetha(a)vt.edu <aswetha(a)vt.edu>* Thanks, Swetha Annavarapu MS in Computer Science Virginia Tech

1 0

June 2023 Board of Visitors Meeting
by frantzme＠vt.edu 13 Jun '23

13 Jun '23

To CS Graduate Students, As a senator and a representative of the CS Grad School on GPSS, please find some highlights from the Board of Visitors (BoV) meeting that closed earlier today. Unfortunately I was unable to make the meeting so all of these notes are abridged or copied directly from the GPSS update. If there are any links missing or you would like to know more information, please email me, Wesley Woo, or the GPSS directly. BoV Update There are new undergraduate scholarships to make a Hokie education more affordable. About $4.6 million dollars in athletics equipment has gone missing. Vice-Rector Ed Baine was unanimously elected to be the next Rector of the BoV. All stipend step amounts will be increased by 5% per state mandate on August 10th, 2023. The new minimum stipend for graduate students will be $2,420 per month over nine months for a total of $21,780 . It is important to note that this is only 66% of the Blacksburg living wage according to the GPSS Task Force Report. Administrators. are working on an implementation plan for the rest of the task force recommendations. VT will receive an additional $40.6 million in revenue from the increase in tuition and fees (unknown % taken from grad students). The BoV and Virginia Tech administrators are not committed to covering comprehensive fees this fall. When BoV Graduate Representative Anna Buhle asked what graduate students should do to cover the $11,000 gap between the new minimum and a living wage, she did not receive a reply. Please find the Cardinal news article here: https://cardinalnews.org/2023/06/06/virginia-tech-to-raise-graduate-student… Please find the VTX news article here: https://news.vt.edu/articles/2023/06/bov-overview-june-2023.html Please find the GPSS Task Force Report here: https://www.provost.vt.edu/content/dam/provost_vt_edu/Graduate%20Assistants… Please find the BoV open session document here: https://bov.vt.edu/assets/Materials-June%205-6,%202023-11.pdf Please find the links to GPSS pages listed below: GPSS Main Page Resolutions Meeting Minutes Meeting Calendar Zoom Meeting Link

1 0