Guanyang Wang
@GuanyangW
Followers
478
Following
327
Media
7
Statuses
137
Assistant professor of statistics @RutgersU, former Ph.D of math @Stanford, undergrad of math @USTC.
Joined July 2020
Why it matters:.1. Training-free boost to image diversity, .2. Up to 100× speed-up for estimating downstream statistics,.3. Reliable uncertainty quantification for assessing diffusion-posterior samplers.
0
1
1
This finding seems universal as it holds across architectures, datasets, sampling schedules, and both unconditional and text-to-image tasks. Perhaps surprisingly, our explanatory experiments show that the score network in diffusion models learns an almost odd function!
1
1
1
Diffusion models start as random Gaussian noise that becomes an image. In our new work w/ Sifan Liu, Bowen Song, Wei Yuan & Liyue Shen, we study that noise itself: flip it and you instantly get a second image that’s significantly anticorrelated—every time.
1
0
0
🚀 Excited to share our newest paper, led by Jing Jia, on #DiffusionModels! Pair the initial Gaussian noise with its negative → two "opposite" images!.📄Paper: .🌐Project: .📝Blog:
1
0
4
RT @sp_monte_carlo: In the interim, I wanted to advertise our YouTube channel - - which contains recordings for the….
0
6
0
RT @sp_monte_carlo: Starting from last October, we (@OnlineMCSeminar, have been running an online seminar on all a….
0
16
0
our last seminar of this semester by Alain Durmus!.
Final Monte Carlo Seminar of the semester! Talk by Alain Durmus on Diffusion models!. 🕰 May 27 11:30 am ET | 8:30 am PT | 4:30 pm London | 5:30 pm Paris | 11:30 pm Beijing.🔗 Join us on
0
0
5
Our first talk on manifold MCMC!.
🗓️ Tue, May 20.🕒 11:30 AM ET / 8:30 AM PT.🎙️ Vivekananda Roy (Iowa State University).📖 “A Riemannian manifold approach to informed MCMC sampling”.🔗 Zoom: Explore how Fisher–Rao geometry yields smarter Metropolis–Hastings proposals!.
0
1
8
Looking forward！.
Next Talk!!.📅 Tuesday, May 13, 2025.⏰ 8:30 am PT | 11:30 am ET | 4:30 pm London | 5:30 pm Paris | 11:30 pm Beijing.Speaker: Ning Ning (Texas A&M).Title: “Convergence of Dirichlet forms for MCMC optimal scaling with dependent target distributions on large graphs”.
0
1
4
RT @OnlineMCSeminar: Next Talk!!.📅 Tuesday, May 13, 2025.⏰ 8:30 am PT | 11:30 am ET | 4:30 pm London | 5:30 pm Paris | 11:30 pm Beijing.Spe….
0
2
0
RT @royhan0611: We performed some explicit computations for the training dynamics of gradient descent in what we call the “finite width pro….
0
3
0
highly recommend!.
📢 Tue, May 6, 2025 • 8:30 PT / 11:30 ET / 16:30 BST / 17:30 CEST / 23:30 CST.Quan Zhou (Texas A&M) will talk about Sampling complexity in high‑dim structure learning. Zoom: Check our Youtube for past talks!.
0
1
3
RT @OnlineMCSeminar: 🎙️ Talk on April 29, 2025: Yuexi Wang (UIUC) ."Optimal Transport-Based Generative Models for Bayesian Posterior Sampli….
0
3
0
Don’t miss our next talk if you like to know more about flow-based generative models!.
📢 Next Talk on Rectified Flow!. Speaker: Qiang Liu (UT Austin) 🗓️ April 22nd ⏰ 8:30 am PT | 11:30 am ET | 4:30 pm London | 5:30 pm Paris | 11:30 pm Beijing. Zoom: 🔍 More details:
0
0
3
RT @OnlineMCSeminar: 🎙️ Monte Carlo Online Seminar.📅 Tuesday, April 15, 2025.🕣 8:30am PT | 11:30am ET | 4:30pm London | 5:30pm Paris | 11:3….
us06web.zoom.us
Zoom is the leader in modern enterprise cloud communications.
0
3
0
MCMC for ranking data!.
Tuesday, April 8, 2025.Speaker: Chenyang Zhong (Columbia).Zoom: Title: A hit and run approach for sampling and analyzing ranking models.Time: 8:30am PT | 11:30am ET | 4:30pm London | 5:30pm Paris | 11:30pm Beijing.Mailinglist:
0
1
4
RT @OnlineMCSeminar: 📢 Upcoming Talk Alert!.🎤 Yuchen Wu (UPenn).📌 Title: Theoretical Advances in Diffusion Models.🗓️ When: April 1st 8:30….
youtube.com
0
4
0
RT @OnlineMCSeminar: 📢Join our next talk by Saif Syed: Scalable sampling of multi-modal distributions using sequential Monte Carlo samplers….
youtube.com
0
5
0