Ray Bai
@raybai07
Followers
1,305
Following
1,049
Media
167
Statuses
839
Assistant Professor at @UofSCStatistics , French bulldog owner, and pop culture enthusiast 👨🏫📊 📚👨🍳🏳️🌈
Columbia, SC
Joined November 2019
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
#今月描いた絵を晒そう
• 180721 Tweets
Mombasa
• 73780 Tweets
Griezmann
• 57232 Tweets
#JusticeForBruce
• 55307 Tweets
#CDTVライブライブ
• 44882 Tweets
PASTEL FRESHTORY X GF
• 25737 Tweets
OUROAD GRAND OPEN HOUSE
• 18731 Tweets
Grizou
• 17820 Tweets
おはスタ
• 17412 Tweets
シャウエッセン
• 16084 Tweets
#chibalotte
• 15190 Tweets
Deschamps
• 14657 Tweets
フルサイズ
• 13662 Tweets
OHMLENG IDEA LIVE
• 11390 Tweets
Finally got a chance to put together all my notes from my stochastic processes class into one single packet. Basically wrote a textbook (305 pages total). 👀😱
13
57
474
Class notes for my grad Stochastic Processes class this semester are all finished and have been put together as a packet! Just 4 more lectures on RL and then looking forward to student project presentations! 👍
#AssistantProfessor
#StochasticProcesses
#statistics
#MachineLearning
7
25
238
My students' class projects have been EXCELLENT! Below is a sample of one of them. The students used curriculum reinforcement learning to train a self-driving car to drive as fast as possible without colliding into other vehicles!
5
9
90
Finally got a chance to put together all my undergrad STAT 517 notes into a single course notes packet -- 233 pages of notes covering generalized linear models (GLMs), random/mixed effects models, & nonparametric/semiparametric models! Very pleased with how the class turned out!
3
4
81
Today is a very special day for me b/c my student Shijie passed his PhD defense! This is my first PhD student ever to defend his diss. Shijie has been a joy to work with (3 papers accepted/published, 4th one about to be submitted). I'm very proud of him!
#phdadvisor
#PHinallyDone
3
1
59
Did you know that there are *two* statistics professors named Ray Bai?
Left: University of Georgia prof Ray Bai who is a probabilist (in a stats department). Right: University of South Carolina prof Ray Bai who is a statistician.
3
1
57
I got my copy of "Handbook of Bayesian Variable Selection"! Lots of great stuff on spike-and-slab priors, global-local shrinkage priors, tree-based models, graphical models, etc. Highly recommended.
0
6
56
When I was a PhD student, my research consisted mainly of coming up with a model first for a “broad” situation (e.g. regression with p>n), deriving nice theoretical properties, and then finding a data set to apply the method to. But now I find it hard to be inspired this way. 1/5
2
5
51
My STAT 718 students did *EXCELLENT* job! They coded SGD/SVRG for SVM, various LASSO algorithms, and EM algorithm for Gaussian mixture model from scratch! They also implemented GANs/WGANs to generate fake images (quality not bad after 10 epochs)!
#statistics
#machinelearning
#AI
0
3
48
Two statistics professors named Ray Bai standing in front of the iconic
@universityofga
Arch
#iconic
#probability
#statistics
#whataretheodds
#verylow
0
2
46
Had an awesome time at Ed George 70 to celebrate the birthday and achievements of esteemed statistician Ed George. Such an inspiration!
0
2
44
Dirichlet processes are really awesome, but this was *byfar* the most challenging topic for me to teach in my teaching career. Pulled it off though! My students also successfully coded DP mixtures for clustering/density estimation from scratch on their final homework! 👍
2
1
42
The final schedule of group project presentations for my Stochastic Processes class! What a diverse and interesting set of topics!😍
#Statistics
#DataScience
2
4
41
My paper "Spike-and-slab group lassos for grouped regression and sparse generalized additive models" (with
@gemma_e_moran
, Joseph Antonelli,
@chenyong1203
,
@MaryRBoland
) has been published in Journal of the American Statistical Association!
Read here:
2
3
39
According to Sir David Cox's mathematical genealogy website, his span of supervising Ph.D. dissertations lasted 60 years! His first PhD student finished in 1954 and his last one finished in 2014. 👀😮 This has got to be a record of some kind.
0
2
38
Coming this December from Chapman and Hall/CRC: "The Handbook of Bayesian Variable Selection." So many great chapters, including one I co-wrote, and many topics covered (e.g. causal inference, graphical models, nonparametrics, etc.)!
3
7
37
Don't doubt yourself, esp not because of age! If you want to go back to school or start new career in your late 20s, 30s, 40s, or beyond, then more power to you! Started my PhD at age 29, which was the best time for me personally b/c I was more mature and self-disciplined by then
5
2
36
(1/2) On the TT academic job market this upcoming year? Here is my blog post about my experience navigating the market for Statistics/Biostatistics TT jobs at research universities in the U.S.A.
1
4
35
In exactly two weeks (May 30), my PhD student Shijie Wang will defend his PhD dissertation "New Deep Learning Approaches to Classical Statistical Problems." This is my first PhD student to defend his diss! Right now we're also preparing Shijie's fourth PhD paper for submission!👍
1
1
35
(1/2) Stan is so good. I wanted to fit a random effects model w/ a more "robust" distribution for the random effect (one that allows heavy tail and/or skewness), and it was seamless. Might have to assign a problem or two on using Stan for my STAT 721 class next semester!
1
0
33
New blog post! "Demystifying Graduate Admissions for Statistics PhD Programs." Hopefully this is a useful resource for folks who are applying to PhD programs. Read below. 👇
0
7
35
Happy to have received an ASPIRE-I grant to develop new methods and algorithms for scalable Bayesian survival analysis! Hoping to use preliminary results to put together an NIH grant.
8
0
34
I am very proud to have been named one of 9 McClausland Fellows by
@UofSC_CAS
for the year 2023! I'm extremely grateful to
@uofscstatistics
and
@UofSC_CAS
for all of their support + all of the wonderful colleagues and students that I get to work with on a daily basis!
We are proud to announce that 9⃣ arts and sciences faculty have been named McCausland Fellows for 2023.
Full story ↪️:
1
4
18
5
1
33
Exactly 10 years ago, I earned my Masters degree in Applied Mathematics from UMass Amherst! This was the catalyst for me eventually becoming a professor! How time flies!
0
0
33
Done preparing for my graduate topics class this semester! Over 400 pages of notes prepared on supervised learning, numerical optimization, unsupervised learning, and deep learning!
#assistantprofessor
#statistics
#MachineLearning
2
3
32
Gave the final talk in one of the final sessions at JSM 2023 (today 10:30-12:20) on Bayesian modeling. What a great group of researchers! Till next year!
0
0
32
I'm also having lunch tomorrow with another statistics professor who is also named Ray Bai (there is another Ray Bai in the UGA Stats Department). An iconic lunch with two Ray Bais! 😂
4
2
31
I always get mildly excited when I get to use something I learned a long time ago in my probability/stat theory courses. In the two most recent revisions I worked on, I ended up using the Borel-Cantelli Lemma and Neyman-Pearson tests in the proofs.
1
1
31
(1/2) My PhD student Zile Zhao will be doing his dissertation proposal in less than a month! Two chapters from his dissertation are in process of being converted to papers, should be available by January. My first time working on survival analysis, it was fun to learn with him!
1
0
28
My paper "Scalable high-dimensional Bayesian varying coefficient models with unknown within-subject covariance" (with
@chenyong1203
and
@MaryRBoland
) has been accepted for publication in Journal of Machine Learning Research!
Read the paper here👉
1
1
28
Watching the Winter Olympics, and I had no idea that star figure skater Nathan Chen was majoring in Statistics and Data Science at Yale! This makes me love him even more. 😍😍😍
#WinterOlympics
#statistics
1
2
27
It's my and my mom's bday month (my mom's bday is on the 9th and mine is on the 26th)! My parents are going back to China next week and I'm going back to SC tomorrow, so we celebrated our birthdays early w/ meal at a Sichuan restaurant and cake. Had a great visit back to Boston!
1
0
28
Congratulations to my PhD student Shijie Wang for successfully passing his PhD proposal, "New Deep Learning Approaches to Classical Statistical Problems"! He did so well and passed with flying colors! 🎊🥳 Grateful to work w/ excellent students
0
0
27
Had a great time presenting some recent work on generative quantile regression at Joint Statistical Meetings this year!
#jsm2022
#statistics
#statistician
#machinelearning
#deeplearning
2
2
28
(1/2) New preprint! "Sparse high-dimensional linear mixed modeling with a partitioned empirical Bayes ECM algorithm." With
@ZgodicAnja
,
@AlexanderMcLai2
, and Jiajia Zhang. We develop a new approach to linear mixed models (LMMs). Read here👉:
1
8
27
Excited to teach STAT 721: Stochastic Processes again in Spring 2024! I get to teach some of my fav topics - Gaussian processes, Dirichlet processes, MCMC, reinforcement learning, & point processes! I enjoy helping students w/ their coding as well.
0
1
27
Had a great time at
@uofscstatistics
end-of-year picnic and award ceremony today! My PhD student Shijie won the Graduate Student Research Award for his outstanding research!
The most rewarding thing about an academic career is seeing your students succeed!
#academia
#professor
1
1
27
First time seeing my parents in nearly 2.5 years! Due to COVID restrictions, it was tough for them to come back to the U.S.A. from China. But they made it this summer! We had a delicious feast at Dumpling House!
#family
#parents
#soupdumplings
#chinesefood
2
0
26
RcppArmadillo is da bomb 💯. By using RccpArmadillo, the average runtime for my student's code decreased from 30 mins (using just R) to less than 2 mins (R integrated with C++) for estimating 600,000 parameters in a 1000 x 600 matrix!
3
1
25
My students did GREAT on their Gaussian processes homework! 😍I made them code exact GP and scalable sparse GP from scratch and use their code to do data analysis. They really rose to the challenge! 1/2
4
1
25
Just finished writing a paper for an epidemiology journal. I found it to be a very rewarding experience since the paper was much less "mathy" than I'm used to writing. I learned so much about communicating statistical ideas + analyses to a non-stat/public health audience!
1
0
23
Had a great time presenting my work on Bayesian mixed-type multivariate regression at the SRCOS 2024 Summer Research Conference!
Tomorrow I'm headed to Niagara Falls, ON, Canada for the
@ICSA_Statistics
Canada Chapter 2024 Symposium where I will again present this work!
0
2
22
I am giving a seminar at
@UMNStatistics
tomorrow on using BART (Bayesian decision tree ensembles) for varying coefficient models (11:30 am-12:30 pm CT in Bruininks Hall Rm 220). This talk also gives a broad overview of tree-based models in general. See you there!
1
1
23
So excited that I finally received a copy of the Handbook of Bayesian Variable Selection in the mail! GREAT job
@MarinaVannucci3
@stephanievdpas
@FJavierRubio1
@fra_stingo
@francescadomin8
@happy_dora1
and others!
1
1
22
Hamiltonian Monte Carlo (HMC) is pretty fascinating... I'd have never thought to use molecular dynamics to make MCMC exploration of the target more efficient! Teaching HMC today in my grad Stochastic Processes class, & my students get to experiment w/ HMC in Stan on their HW!
1
0
22
Congratulations to my PhD student Shijie Wang! His first paper "Generative Multiple-purpose Sampler for Weighted M-estimation" has been accepted for publication by the Journal of Computational and Graphical Statistics! Well done, Shijie! 🎉🥳
0
0
23
Preparing my notes on Metropolis-Hastings algorithm for next week's lectures, including proof why it works, and it truly is a spectacular result! CiSE was correct to include MH on its list of top 10 algorithms of the 20th century. Right up there w/ fast Fourier transform.
2
1
22
@timgill924
Sunk cost fallacy is very real. Also why some continue to languish in postdocs/VAPs/temporary positions for many years after (and ea. additional year becoming even less competitive on TT market). PhD and postdoc supervisors should be forthright w/ supervisees about their chances
2
1
19
I'll be giving a talk at EcoSta in Tokyo next Aug 2023 in an invited session on Bayesian structure discovery! I'll present some joint work w/ PhD student collaborator
@rh8liuqy
and Xianzheng Huang on Bayesian modal regression! This paper should be out in the next few weeks! (1/2)
3
1
20
(1/2) The newest draft of my paper on high-dimensional Bayesian varying coefficients is now on arXiv! Title is now "Scalable high-dimensional Bayesian varying coefficient models with unknown within-subject covariance." Read it here if interested:
2
2
20
Had a great time chairing and acting as the discussant for this session at
#ISBA2022
yesterday! Great talks about:
1) extending BART for density regression
2) using multi-outcome BART for Pareto optimization
3) a new tree prior to better handle splits on categorical predictors
Off to Montreal, Canada for ISBA2022! Consider joining the parallel session "Junior advances in Bayesian treed regression" that I am chairing on Tuesday at 1:30 pm!
0
0
6
1
2
20
(1/2) Met my undergrad student Hung-Tien (BS '23) for coffee at Knowledge Park Cofffee Co. He'll be starting his PhD in Computer Science at UNC-Chapel Hill this fall!
1
1
20
(1/2) Went through the proof that an MCMC chain produced by the Metropolis-Hastings algorithm has the target distribution p(x) as its stationary distribution in my grad Stochastic Processes class yesterday. A very powerful result! One of my fav proofs in the class.
1
0
18
Celebrated our 6-year anniversary at Di Vino Rosso last night with excellent cocktails, a charcuterie board, and gnocchi and scallops! ❤️❤️ Highly recommend this place for those in/passing through Columbia, SC!
#cocktails
#italianfood
#nomnomnom
2
0
19
My students' projects: 1 project on Dirichlet process, 2 projects on reinforcement learning, 2 projects on option pricing, 1 project on numerical SDEs, 2 projects on Gaussian process, 1 project on hidden Markov models, 1 project on Bayesian additive regression trees.
GREAT job!
2
0
19
Excited to be teaching an undergrad class for the first time in Fall 2022! Teaching STAT 517: Advanced Statistical Models. I will be covering GLMs, semiparametric/nonparametric models, and spatial models!😁
0
2
19
@PFort35
I thought
@ta_broderick
's 3-part tutorial series on Bayesian nonparametrics gives a nice introduction + intuition of the Dirichlet process:
Pt 1:
Pt 2:
Pt 3:
2
3
18
The slides for my PhD student's final dissertation defense this Thursday are ready! 😁 Let's gooo
2
2
20
Apartment is all set up for the arrival of our new puppy Ollie, who will be joining us tomorrow! I'm filled with anticipation and excitement!
#frenchbulldog
#doggydaddy
😍🐶
0
0
19
Writing my final set of lecture notes for my class (notes on Dirichlet process mixtures) and a bit sad... but it was very rewarding to teach this class! I wrote ≈300 pages of notes on stochastic modeling over the past 3 months. Wonder if I should try to monetize it, lol. 1/2
1
0
19
2023 was not too bad a year for me.
- 1 new paper published in JMLR
- 2 PhD students completed their PhD proposals & will graduate this year
- submitted a new NSF proposal
- updated my high-dimensional data grad elective class to cover deep learning and deep generative models.
0
0
18
I'm giving a talk at UNC-Chapel Hill tomorrow (Mon., Mar. 4) from 3:30-4:30 pm in Hanes 120! I'll be talking about my work that just got accepted by JCGS (joint w/ Shijie Wang and Minsuk Shin) -- generative quantile regression with a variability penalty!
0
0
18
So excited to be giving a talk at University of Glasgow School of Mathematics & Statistics in Scotland in December!
I have never been to Scotland before, and I may pair this visit w/ some personal travel. Any recommendations on fun things to see and do in Scotland?
6
0
18
Are there any particular nuances that should be addressed in a statistical model for this data? Approaching statistics this way can usually lead to methods AND (sometimes) theory. So basically, instead of theory->method->data, I now prefer data->method->theory. 5/5
1
0
18
Great thing about teaching graduate courses is the amount of freedom you get to design your class. Decided to make my grad-level Stochastic Processes class focused heavily on Bayesian computing (MCMC) and (applied) Bayesian nonparametrics, /1
2
1
18
(1/2) Coming soon - newest version of my paper on varying coefficient models! At reviewer's suggestion, the theoretical results will be moved to a separate manuscript. Instead, this updated draft focuses just on methodology & computation -- now with uncertainty quantification!
1
1
17
Great time visiting Clemson University School of Mathematical and Statistical Sciences today! My first non-virtual, in-person department seminar this year!
Looking forward to giving my next in-person colloquium talk next month at Virginia Tech Department of Statistics!
0
0
18
(1/2) So exciting! One of my PhD students will be doing his PhD proposal "New Deep Learning Approaches to Classical Statistical Problems" in January 2023 (final defense will be in Spring 2024 if all goes well)!
My very first PhD proposal as a supervisor and not as a student. 😀
2
2
17
I've been seeing lots of posts about the (faculty) academic job market so I thought I'd pull up an old blog post about my personal experience on it:
Mainly concerns jobs at research universities-hope it helps some folks out
#academictwitter
#academicjobs
2
3
18
One week from today, my PhD student Shijie will defend his dissertation! This is my first PhD student to defend his diss, so this is an especially memorable milestone in my career. Shijie has also accepted a job offer at an AI company out in CA!
#AssistantProfessor
#ProudAdvisor
0
0
18
The "hierarchy principle" - that if interactions are included in the model, then the corresponding main effects must also be included - is not always sensible. In fact, in my spike-and-slab group lasso paper, we did not enforce hierarchy when including nonlinear interaction terms
New paper
On the common belief that interactions and non-linear terms should always be accompanied by their “main effects”. Some arguments are given why that’s not always the case
Excellently led by
@tmorris_mrc
, w/
@t_mpham
10
100
394
0
0
18
My new paper "Neural-g: A Deep Learning Framework for Mixing Density Estimation" (with my excellent former student Shijie Wang and our collaborators Saptarshi Chakraborty and Qian Qin) is now available on arXiv!
Read the preprint here if interested:
1
2
19
Can't express enough how much of an absolute joy it is to teach Stochastic Processes this semester. Decided to finish the semester with teaching the students how to parallelize HMC in Stan and (time permitting) reinforcement learning!
3
0
18
Excited to teach reinforcement learning and MCMC after the break! I think RL and MCMC are great ways to illustrate key concepts from Markov chains and get students excited about them!
1
0
18
Check out the *awesome* group projects that the students in my STAT 721: Stochastic Processes class are doing! I'm very much looking forward to their presentations at the end of the semester! Love working with smart and motivated students.
#AssistantProfessor
#statistics
3
2
17
Done writing a grant application! Hope that I get it so I can support one of my PhD students as a Research Assistant in the 2023-2024 academic year. 🤞🤞 Now I can relax a bit the next couple days!
Looking forward to the ASA-South Carolina Palmetto Symposium tomorrow!!!
0
0
16
First draft of my paper on joint quantile estimation w/ my PhD student Shijie Wang and Dr. Minsuk Shin is complete! 😁We are working hard to polish this paper and hope to have it posted on arXiv in the next few weeks!
0
1
17
The past 6 weeks were the most stressful I've experienced as an Asst Prof. And I didn't even teach! But on the plus side,
✔️Major revision done and resubmitted.
✔️PhD student finished his written and oral proposal.
✔️Finished and submitted a new NSF grant!
Time for a break.
2
0
17
(1/2) Very excited about the new and much improved version of our paper which we will be posting (hopefully) later this month! We introduce a Bayesian approach for conducting variable selection in correlated, mixed response (both discrete and continuous) multivariate regression!
1
2
16
(1/2) So exciting when you find out that your student's research has made a great breakthrough! Found out that my PhD student's implementation of a Dirichlet process mixture model (DPMM) not only works in Stan, but the parallelized divide-and-conquer version of it *ALSO* works!
2
0
16
Trying to work on a grant proposal, but these two just want me to play with them. How can I resist those faces?!
#frenchbulldogs
#frenchies
#frenchieowner
1
1
17
I'll be at the 2022 ICSA Applied Statistics Symposium in Gainesville, FL!
I'm giving a talk in Session 5D: Statistical Methods For Complex And High Dimensional Data (Tue June 21, 10:00-11:40 am), along with Qian Qin, Xuan Cao, and Jairo Fúquene-Patiño! Come check us out!
1
0
17
I actually have an easier time coming up with ideas now if I look at a real data set FIRST, determine what scientific questions are of interest from this data, and what challenges exist for analyzing this data. 2/5
1
2
17
Come learn about flexBART + other exciting methodological developments about BART in our session "Junior advances in Bayesian treed regression" at
#ISBA2022
tomorrow at 1:30 pm!
I'm super excited to attend
#ISBA2022
in Montreal next week and then EURO2022 in Espoo! I'll be talking about some fun I've been having w/ BART over the last few years. (1/17)
2
2
28
0
0
17
My PhD advisor Malay Ghosh gave the Deming Lecture at JSM yesterday! I also got a chance to catch up with him a bit earlier in the day. Always a pleasure!
0
0
17
I'm speaking in sessions at the upcoming conferences! Let me know if you'll be there :)
UP-STAT 2022 (Buffalo, NY May 2-4)
ICSA 2022 (Gainesville, FL, Jun 19-22)
ISBA 2022 (Montreal, Canada, Jun 26-Jul 1)
EURO 2022 (Espoo, Finland, Jul 3-6)
JSM 2022 (Washington, D.C., Aug 6-11)
3
0
16
It's strawberry season in South Carolina! Picked some juicy, delicious strawberries at Cottle Strawberry Farm today! Baking strawberry cheesecake this afternoon. 🤤🍓
1
0
16
Check out these awesome plots made by students in my Stochastic Processes class this past semester!
#statistics
#stochasticprocesses
#spatialstatistics
#mathematicalfinance
#reinforcementlearning
#gaussianprocesses
#mcmc
#dirichletprocesses
0
0
16
Congrats to my PhD student Shijie Wang for winning a Citizenship award at the
@uofscstatistics
awards picnic today. Last year, Shijie won the James D. Lynch Graduate Student Research Award
I was also very touched by his farewell speech (graduating this summer)!
#proudPhDadvisor
0
0
16
Great news! Application fee for the upcoming graduate application cycle has been WAIVED at U. of South Carolina!
Prospective Stat grad students: Consider applying to
@uofscstatistics
PhD and MS programs in Statistics! It's free to apply! Apply by Jan 15
2
3
15
(1/2) Newest draft of my paper is finally ready! "Bayesian Group Regularization in Generalized Linear Models with a Continuous Spike-and-Slab Prior"!
Read it here if you'd like:👉👉
2
0
16
Had an excellent visit to UNC-Chapel Hill Dept of Statistics and Operations Research today! Got a nice tour, met with faculty and grad students -- many of whom asked excellent questions during my talk! Glad to talk about conditional density estimation and deep generative models!
0
0
15
1/2 Got a chance to browse through the final proof of "The Handbook of Bayesian Variable Selection" (Eds. Mahlet Tadesse and
@MarinaVannucci3
). It looks EXCELLENT! A lot of great stuff on methodology, computation, and theory. Something for everybody!
2
1
16
Prepping my notes for the final unit of my class (≈5 lectures on the Dirichlet process), and I came across this lovely paper by Jeff Miller: "An elementary derivation of the Chinese restaurant process from Sethuraman’s stick-breaking process." 1/2
1
1
16
New blog post! "Reflecting on the first two years on the tenure-track: Part I." I discuss what I learned about teaching and supervising PhD students these past 2 years. Read here👇
0
0
15