yingzhen

@liyzhen2

teaching machines 🤖 to learn 🔍 and fantasize 🪄 now 🇬🇧 @ImperialCollege @ICComputing 🇸🇬 @NTUsg @NTU_ccds ex 🇬🇧 @MSFTResearch @CambridgeMLG

yingzhenli.net

Joined April 2012

164 Following

4,448 Followers

Replies Reposts Media

yingzhen
@liyzhen2
Sep 25
This is now also accepted at #NeurIPS2026
yingzhen
@liyzhen2
May 11
Our own answer: structured coupling arxiv.org/abs/2605.07676 - flow matching with VAE-based coupling - VAE encoder & flow sharing networks - VAE decoder init. + flow refinement for sampling flow matching 🤝 VAEs -> good representation & sample quality🚀
yingzhen
@liyzhen2
Sep 9
In AGI(?) era maybe the classic debate “math is discovery and/or invention” is worth for a refreshing discussion
yingzhen
@liyzhen2
Jun 6
Excited to join @NTUsg , will split time between @NTU_ccds and @ICComputing Let’s combine the strengths 💪 of AI R&D in 🇸🇬🇨🇳 and 🇬🇧🇪🇺 for scaling up probabilistic ML 🚀 Reach out for PhD & postdoc if interested 🤗 Any cool ML researchers in 🇸🇬 that I should reach out?
9
yingzhen
@liyzhen2
Jun 3
@naoki_kiyohara did thought about strong solution when working on neural stochastic flows arxiv.org/abs/2510.25769, but we didn't know how to enforce consistency on a stochastic path. The Legendre polynomial idea is wise 👍 we've tried related ideas on GPs arxiv.org/abs/2502.08736
Niklas Rindtorff
@Niklas_TR
Jun 2
Introducing Strong Stochastic Flow Maps TLDR: Stochastic Flow Maps where we learn the stochastic solution path. Work led by Sam McCallum, @zwblasingame, with Timothy Herschelll, @AlexanderTong7, and @JamesFosterBath Arxiv: arxiv.org/pdf/2606.01086 Code: github.com/sammccallum/ss…
GIF
yingzhen
@liyzhen2
May 30
found conf. books 💎of the winter AI party conf ~10yrs ago crazy speaker list in 2015: - 4 Turing awardees, Dean, Vinyals, Dally giving tutorials - Tibshirani/Vapnik/Ghahramani among invited speakers hopefully this year's speakers will be equally exciting 😊