Jack Qing

[Tiange Xiang @xxtiange]

Reconstructing occluded humans from monocular video can be nice and fast! 🎆 I’m excited to share our new paper “OccFusion: Rendering Occluded Humans with Generative Diffusion Priors” 🧵
📖https://t.co/HFuFvpkVaA
🌐https://t.co/WXCsOg6H8A

[Ken Liu @kenziyuliu]

The idea of "machine unlearning" is getting attention lately. Been thinking a lot about it recently and decided to write a long post: https://t.co/YWFco5xNNq 📰

Unlearning is no longer just about privacy and right-to-be-forgotten since foundation models. I hope to give a gentle overview of unlearning and touch on things like copyright, NYT v. OpenAI, NeurIPS unlearning challenge, retrieval-based systems, AI safety, & pretending to unlearn.

I hope it'll be a fun weekend read!

[Tiange Xiang @xxtiange]

Human NeRFs have shown incredible abilities rendering 3D avatar from single monocular videos. But can they train on videos with occlusions? I’m excited to share our new paper “Wild2Avatar: Rendering Humans Behind Occlusions”🧵
📖https://t.co/kr7MeRFyIs
🌐https://t.co/19hZtslhpv

[Jack Qing @QingJack]

@IlyaKuzovkin @ZijiaoC @HelenJuanZhou @TarrLab @humanscotti @iScienceLuvr @NishimotoShinji but I don’t think decoding a rich text that describes the visual sitmulus well is an easier task than reconstructing images directly.

[Jack Qing @QingJack]

@IlyaKuzovkin @ZijiaoC @HelenJuanZhou @TarrLab @humanscotti @iScienceLuvr @NishimotoShinji I think @HelenJuanZhou and @humanscotti alreadly discussed this issue quite clear, but I also like to add something. I agree that the reconstrction should ideally be exactly the same as the visual stimulus (if we can excluding the mind wandering and other artifacts),

[Jack Qing @QingJack]

@IlyaKuzovkin @ZijiaoC @HelenJuanZhou @TarrLab @humanscotti @iScienceLuvr @NishimotoShinji decoding text and decoding images are two different tasks (involving different brain regions and different brain encoding model). If you mean decoding text from visual cortex first and use that to generate images, that’s definitely a valid method.

[Jack Qing @QingJack]

@IlyaKuzovkin @ZijiaoC @HelenJuanZhou @TarrLab @humanscotti @iScienceLuvr @NishimotoShinji And it’s a tradeoff when we are limited by current methods and data. I believe as our algorithms improved, we can achieve this ultimalte goal (we already see a lot great work like the one from @humanscotti )

Regarding ur text decoding comments, correct me if I misunderstood,

[Jack Qing @QingJack]

@IlyaKuzovkin @ZijiaoC @HelenJuanZhou @TarrLab @humanscotti @iScienceLuvr @NishimotoShinji which is why most the recon papers report different metrics (both semantic and pixel level) and that is the ultimate goal of this field of research. Targeting pixel level or semantic level first are trying to approach this ultimate goal from two different directions.

[Jack Qing @QingJack]

@negar_rz https://t.co/u27acXT2Ju

[Shaonan @ShaonanWang]

We will be hosting a workshop at AAAI this year, and we are currently accepting paper submissions on the captivating theme of ‘Brain Encoding and Decoding.’ We warmly invite you to submit your work and join us at AAAI to engage in meaningful discussions about this topic.

[Jack Qing @QingJack]

Thrilled to share that our work has been accepted by #NeurIPS2023 as an Oral presentation!!

Project page: https://t.co/JyXA2k41dL

We've cracked the code to decode photorealistic videos from human brain recordings! 📽️➡️🧠

@_akhaliq @NeurIPSConf
#deeplearning #MachineLearning

[AK @_akhaliq]

Cinematic Mindscapes: High-quality Video Reconstruction from Brain Activity

propose Mind-Video that learns spatiotemporal information from continuous fMRI data of the cerebral cortex progressively through masked brain modeling, multimodal contrastive learning with spatiotemporal attention, and co-training with an augmented Stable Diffusion model that incorporates network temporal inflation

paper page: https://t.co/XzbqItFtN1

[Zijiao Chen @ZijiaoC]

🧠✨Super excited to share that our work Mind-Video has been accepted by #NeurIPS2023 as an Oral presentation! 📍 Catch us live in New Orleans!

TL; DR: We've cracked the code to decode photorealistic videos straight from human brain recordings! 📽️➡️🧠
@NeurIPSConf #AI4Science

[AK @_akhaliq]

Cinematic Mindscapes: High-quality Video Reconstruction from Brain Activity

propose Mind-Video that learns spatiotemporal information from continuous fMRI data of the cerebral cortex progressively through masked brain modeling, multimodal contrastive learning with spatiotemporal attention, and co-training with an augmented Stable Diffusion model that incorporates network temporal inflation

paper page: https://t.co/XzbqItFtN1

[Jack Qing @QingJack]

Check out our new research on brain decoding!

[Zijiao Chen @ZijiaoC]

🧵🧠 We're witnessing incredible scientific progress in image & text reconstruction from fMRI nowadays. But what about reconstructing video from fMRI? Allow me to introduce our recent preprint: Mind-Video
https://t.co/VL2KXz8o9K
https://t.co/KyNtsCxDIJ
https://t.co/bhjz0PDlS6

[Zijiao Chen @ZijiaoC]

🧵🧠 We're witnessing incredible scientific progress in image & text reconstruction from fMRI nowadays. But what about reconstructing video from fMRI? Allow me to introduce our recent preprint: Mind-Video
https://t.co/VL2KXz8o9K
https://t.co/KyNtsCxDIJ
https://t.co/bhjz0PDlS6

[NBC Nightly News with Tom Llamas @NBCNightlyNews]

Researchers are developing artificial intelligence technology they say generate the very images in our brains.

@byjacobward shares more on the extraordinary research and the concerns about what mind reading technology could mean for the future: https://t.co/aLxnmbUuJr

[Jack Qing @QingJack]

@ZijiaoC @CVPR Thanks for providing this wonderful opportunity to share our work😊

[Zijiao Chen @ZijiaoC]

Excited to share that MinD-Vis has been accepted by CVPR23. 🥳

TL;DR: We use a two-stage model (self-supervised learning + latent diffusion model) to generate nature images from brain recordings.

Check out our project page: https://t.co/li0smBy911

[University of Sydney @Sydney_Uni]

Last night we celebrated the incredible achievements of our 18,000 graduates of 2020 with a live-streamed celebration event from the Great Hall.

To the graduating class of 2020, congratulations, you did it.

#USYDClassOf2020.