Yujia Huang
@YujiaHuangC
Followers
656
Following
210
Statuses
35
PhD student @ Caltech. Machine learning & deep learning. Generative models. Love music.
Pasadena, CA
Joined September 2017
Unfortunately I could not be at Vienna because of visa issues. But my advisor @yisongyue will give the talk, and my collaborator Adishree Ghatare will present the poster. Don't hesitate to DM or email me if you want to discuss in more details or have questions:) Huge thanks to all other amazing collaborators @lyzfantastic @acbuller @qsh_zh @cmssastry Siddarth Gururani and @osageev!
0
0
7
5/5 Code and checkpoints: Great collaboration with Adishree Ghatare @lyzfantastic @acbuller @qsh_zh @cmssastrySiddharth Gururani @osageev and @yisongyue!
1
0
11
@patrickmineault @brain_score @doristsao @wn8_nie Although it is looks less like predictive coding explicitly, but the diffusion models are very powerful to learn the data distribution. Maybe it would be interesting to explore the connection to PC.
1
0
2
@SaberaTalukder @robbieking1000 The general idea for formal guarantee on robustness is to bound the range of the output of a neural network given input perturbations. There are many ways of doing that, like linear relaxation (the previous beta-crown) paper and Lipschitz bound (e.g. my work).
0
0
2
RT @Deep__AI: Training Certifiably Robust Neural Networks with Efficient Local Lipschitz Bounds by @YujiaHuangC et…
0
2
0
@genintelligent Had a very interesting discussion with @kanjun and @joshalbrecht. I feel the most difficult part for general intelligence research is to find the right questions to ask. What do we actually want our ml model to learn? Still thinking hard about it.
0
1
6
RT @genintelligent: Yujia Huang (@YujiaHuangC) from Caltech discusses neuro-inspired ML this week! We explore: 🏗 A…
0
1
0
The proposed CNN-F introduces generative feedback with latent variables to existing CNN architectures. We enforce self-consistency between feedforward classification and feedback generation. @AnimaAnandkumar @doristsao @ZhidingYu @TanNguyen689
0
0
2
Excited about #NeurIPS2020 next week! Hopefully we can take more advantage from online conference after getting more used it. I plan to check the accepted papers and mark the interesting papers to check out;)
@NVIDIAAI I am most excited or interested about the paper "Neural Networks with Recurrent Generative Feedback" @NVIDIAAI
#NeurIPS2020
1
2
28
RT @yisongyue: I've written out a collection of tips for crafting your faculty application package (mainly for computer science). Distille…
0
54
0