Associate Professor @ECEILLINOIS @IllinoisCS, working on Computer Vision and Machine Learning
Joined February 2018
- Tweets64
- Following175
- Followers1.2K
- Likes3
Congrats to @hkchengrex for impressive work on generating audio from video and/or text.
🔊 Tired of silent Sora videos? Bring them to 𝓵𝓲𝓯𝓮 with MMAudio!
Generate synchronized audio directly from video and/or text inputs.
Code's ready — try it now with one simple command:
github.com/hkchengrex/MMAudi…
Paper is in perp. 🧪
Congrats @ECEILLINOIS, @IllinoisCS students and collaborators for great #NeurIPS2022 work on PAC-Bayesian Bounds, Animal Reconstruction, Invariant and Equivariant Deep Nets, GAN training with limited data and reinforcement learning from demonstrations. Stop by to get to know more
DigGAN: Discriminator gradIent Gap Regularization for GAN Training with Limited Data – Tiantian Fang, @RuoyuSun_UI
Hall J #519, Wed Nov.30, 11:30am - 1pm CST
Paper: openreview.net/pdf?id=azBVn7…
Project: github.com/AilsaF/DigGAN
CEIP: Combining Explicit and Implicit Priors for Reinforcement Learning with Demonstrations - @KaiYan86524445, Yuxiong Wang
Hall J #129, Wed Nov.30, 4:30pm - 6pm CST
Paper: arxiv.org/abs/2210.09496
Project: 289371298.github.io/jekyll/u…
Congrats to @hkchengrex (@IllinoisCS, @HkustSc) on a fun deep net study of the 1968 Atkinson-Shiffrin memory model for video object segmentation. People seem to like the code: github.com/hkchengrex/XMem. What's your favorite psychology/brain science hypothesis used in deep nets?
![GitHub - hkchengrex/XMem: [ECCV 2022] XMem: Long-Term Video Object Segmentation with an Atkinson-...](https://pbs.twimg.com/card_img/2118855586918039552/zxcJFTj6?format=png&name=800x419)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model
abs: arxiv.org/abs/2207.07115
project page: hkchengrex.github.io/XMem/
github: github.com/hkchengrex/XMem
Congrats to (almost) former @ECEILLINOIS student @adamliu123 (ioujenliu.github.io) for successfully graduating with impressive work "Towards Efficient Multi-Agent Deep Reinforcement Learning". Enjoy the ceremonies and keep up the great work.
Tired of training agents by exploring environments randomly? Talk to @adamliu123, @ericxyuan, @Cote_Marc, @pyoudeyer about agents asking for knowledge @icmlconf #ICML2022
Spotlight: 7/21 3:55pmEDT,#307
Poster: 7/21 6pmEDT,Hall E905
arxiv.org/abs/2205.06111
ioujenliu.github.io/AFK/
Congrats @ECEILLINOIS, @IllinoisCS students @RaymondYeh, @YuanTingHu1, @RenZhongzheng, @bowenc0221, @ColinGraber and colleagues for great #CVPR2022 work on total variation, universal & instance segmentation, forecasting and neural volumetric object selection. Talk to them @CVPR.
Curious to know what's happening next? Talk to @ColinGraber, Cyril Jazra, Wenjie Luo, Liangyan Gui about Joint Forecasting of Panoptic Segmentations
Oral: 06/21 1:30pm-3:00pm
Session 1.2, 06/21 2:30pm-5pm (ID 7b)
Paper: arxiv.org/abs/2204.07157
Web: github.com/cgraber/psf-diffa…
Want to know how to select objects in your neural volume (NeRF or MPI)? Talk to @RenZhongzheng, @aseemaa, Bryan Russell, @oliverwang81 about Neural Volumetric Object Selection
Session 2.1, 06/22 10am-12:30pm (ID 87a)
Paper: arxiv.org/abs/2205.14929
Web: jason718.github.io/nvos/
Fun work on neural volumetric object selection with @RenZhongzheng (@IllinoisCS), @aseemaa, Bryan Russell and @oliverwang81 from Adobe.
arXiv: arxiv.org/abs/2205.14929
website: jason718.github.io/nvos/
Alex Schwing retweeted
Excited to share that our new work (w/ @adamliu123 @Cote_Marc @pyoudeyer @alexschwing ) has been accepted to #ICML2022. 🎉🎉🎉
Led by our amazing intern @adamliu123 , we investigate how RL agents can leverage external knowledge sources by querying them using language. [1/6]
Asking for Knowledge: Training RL Agents to Query External Knowledge Using Language
abs: arxiv.org/abs/2205.06111
project page: ioujenliu.github.io/AFK/
Congrats to (almost) former @ECEIllinois, @IllinoisCS students @ColinGraber, @JyotiAneja and also to other recent graduates @UnnatJain2010, @SafaMessaoud3, @RaymondYeh. Enjoy the graduation ceremonies and keep up the great work.
