- nlplab - mailman.cs.vt.edu

next week meeting schedule
by Lifu Huang 29 Nov '20

29 Nov '20

Monday: 11:00-11:30am: Sijia, 11:30-12:00am: Adheesh, 2-2:30pm: Shuaicheng, 6:00-7:00pm: reading group discussion, Sijia lead. Wednesday: 9-10am: Kaiwen Thursday: 1:30-2pm: Sijia 2-2:30pm: Shuaicheng 2:30-3pm: Adheesh Hope everyone can get some scores with your preliminary approaches, then we can resume our weekly group meeting. Lifu

1 0

next week meeting schedule
by Lifu Huang 22 Nov '20

22 Nov '20

Monday is the deadline of NAACL’2021. I have to focus on NAACL submissions and move individual meetings to Tuesday. Monday: 5:30-6:30pm: reading group discussion, Adheesh lead. Tuesday: 11-11:30am: Adheesh 11:30-12pm: Shuaicheng 2:30-3pm: Ali, 4:30-5pm: Sijia Wednesday: 9-10am: Kaiwen Thursday: 1:30-2pm: Sijia 2-2:30pm: Ali 2:30-3pm: Adheesh 3-3:30pm: Shuaicheng Lifu

1 0

next week meeting schedule
by Lifu Huang 15 Nov '20

15 Nov '20

Monday: 10:00-10:30am: Sijia, 10:30-11:00am: Adheesh, 2-2:30pm: Shuaicheng, 2:30-3pm: Ali 5:30-6:30pm: reading group discussion, Shuaicheng lead. (I won’t be able to attend due to a conflict with another meeting) Wednesday: 9-10am: Kaiwen Thursday: 1:30-2pm: Sijia 2-2:30pm: Ali 2:30-3pm: Adheesh 5-5:30pm: Shuaicheng Everyone please also check the related papers to your project from EMNLP2020. Lifu

1 0

next week meeting schedule
by Lifu Huang 08 Nov '20

08 Nov '20

Monday: 11:30-12am: Sijia, 12-12:30am: Adheesh, 2-2:30pm: Shuaicheng, 2:30-3pm: Ali 5:30-6:30pm: reading group discussion, Sijia lead. Wednesday: 9-10am: Kaiwen Thursday: 1:30-2pm: Sijia 2-2:30pm: Ali 2:30-3pm: Adheesh 5-5:30pm: Shuaicheng

1 0

next week meeting schedule
by Lifu Huang 01 Nov '20

01 Nov '20

Sunday: 11-12pm: Kaiwen Monday: 11-11:30am: Sijia, 11:30-12am: Adheesh, 2-2:30pm: Shuaicheng, 2:30-3pm: Ali 5:30-6:30pm: reading group discussion, Ali lead. Wednesday: 9-10am: Kaiwen Thursday: 1:30-2pm: Sijia 2-2:30pm: Ali 2:30-3pm: Adheesh 3:3:30pm: Shuaicheng Lifu

1 0

[NLPLab] next week meeting schedule
by Lifu Huang 25 Oct '20

25 Oct '20

Monday: 10-10:30am, Sijia, check implementation progress 10:30-11am, Adheesh, check updated baseline score and implementation of new idea 2-2:30pm, Shuaicheng, check implementation progress 2:30-3pm, Ali, fake news detection dataset survey 5:30-6:30pm, reading group discussion, Adheesh lead. Wednesday: 9-10am: Kaiwen, dialogue summarization Thursday: 1:30-2pm, Sijia, check implementation progress 2-2:30pm, Ali, fake news detection dataset survey 2:30-3pm, Adheesh, check implementation progress 3-3:30pm, Shuaicheng, check implementation progress Lifu

1 0

[NLPLab] next week meeting schedule
by Lifu Huang 18 Oct '20

18 Oct '20

Monday: 10-10:30am, Sijia, check implementation progress 10:30-11am, Adheesh, get coreference baseline errors done 2-2:30pm, Shuaicheng, check implementation progress 2:30-3pm, Ali, discuss plans for fake news detection survey 5:30-6:30pm, reading group discussion, shuaicheng lead. Wednesday: 9-10am: Kaiwen, dialogue summarization Thursday: 1:30-2pm, Sijia, check implementation progress 2-2:30pm, Ali, discuss papers about fake news detection 2:30-3pm, Adheesh, discuss approaches for joint entity and event coreference 3-3:30pm, Shuaicheng, check implementation progress In your progress slides, please summarize (1) what you have done since the last meeting (2) what you will finish by the next meeting. Thanks, Lifu

1 0

gpu server
by Lifu Huang 15 Oct '20

15 Oct '20

Our gpu server is ready to use. I have created an account for Sijia, Shuaicheng an Adheesh. The user name and password is your pid. The external IP address is: 128.173.236.231 The dns entry is: nlplab1.cs.vt.edu (Both should be active now.) nlplab1 has /home mounted from a NFS server, so we will have 86 TB of disk to play with in /home. I have created a /home/data so that you can put all useful datasets, benchmarks, resources here. When you put a dataset/resource in the server other than your home dir, please add a description here so that we can track the usage of it later: https://docs.google.com/spreadsheets/d/19vy_9ReS9f50vnI5D_N4RKYY4lN920zEtuu… Please also sign-up a gpu here: https://docs.google.com/spreadsheets/d/19vy_9ReS9f50vnI5D_N4RKYY4lN920zEtuu… Each card can be signed by 2-3 students. @Ali, I can also create an account for you if you can send me your VT pid. Let me know if you have any questions and suggestions. Lifu

1 0

Next week meeting schedule
by Lifu Huang 11 Oct '20

11 Oct '20

Monday: 5:30-6:30pm: reading group discussion, Sijia lead. Wednesday: 9-10 am: Kaiwen, on dialog summarization 2:30-3:30: Ali, on potential research topics. 3:30-4:30: Sijia, on QAIE progress 4:40-5:30: Shuaicheng, on temporal relation prediction 5:30-6:30: Adheesh, on error analysis and new ideas for coreference resolution Please send me your progress slides one day before the meeting. Thanks, Lifu

1 0

Fwd: DLRL meeting of Oct. 8 - guest speaker: Dr. Lourentzou: Semi-supervised Learning tutorial
by Lifu Huang 08 Oct '20

08 Oct '20

---------- Forwarded message ---------- From: EdFox <fox(a)vt.edu> Date: Oct 1, 2020, 9:33 PM -0400 To: dlrl-g(a)vt.edu <dlrl-g(a)vt.edu> Cc: EdFox <fox(a)vt.edu>, Lourentzou, Ismini <lourent2(a)illinois.edu> Subject: DLRL meeting of Oct. 8 - guest speaker: Dr. Lourentzou: Semi-supervised Learning tutorial > Hi! > > Please spread the word about the guest speaker presentation at next week’s DLRL meeting. As usual, this will be > Thu 3:30-4:30pm in Zoom (see details below) > Thanks, eaf > > Speaker: Dr. Ismini Lourentzou > (New faculty member in CS; recall her seminar from April: > https://cs.vt.edu/News/Seminars/ilourentzou.html) > > Title: Semi-supervised Learning tutorial > > Abstract: > The recent success of deep learning can be accredited to large annotated datasets and better computational infrastructures available. However, in new domains and applications, acquiring large amounts of data is a costly and time-consuming process. Even with active learning annotation pipelines, typically, most of the data available remain unlabeled. Semi-supervised learning is a class of algorithms between supervised and unsupervised learning, that aims to model the data distribution using both labeled and unlabeled data, in order to train a better performing model than what we could obtain by using only the labeled subset. This talk will cover a broad overview of recent advances in semi-supervised learning in a coherence timeline, from self-training and co-training to FixMatch (NeurIPS 2020). Finally, we will conclude with some limitations and open problems for future research. > > Link: > https://virginiatech.zoom.us/j/94931034669?pwd=U1NYcXF3Sk1XUGdLeHJja1V4bEZt… > > PW: JCDL20 > > Meeting ID: 949 3103 4669 >

1 0