Teo Reu

2nd-year DPhil at Oxford University. Generative Modeling.

[Alicja (Ala) Maksymiuk @AlaMaksymiuk]

🚨We present MacroGuide: the first model to generate arbitrary macrocycles in 3D 🎉

How? Topological guidance with persistent homology!🧵⬇️ [1/6]

With @duplessis_alex @mmbronstein @AlexanderTong7 @fjduarteg @ismaililkanc

[Teo Reu @TeoReu]

Neurips spotlight! If you care about failure cases of Rectified Flows and about what gradient variance reveals about the optimality of transport vector fields, check out https://t.co/7dH73GfCIt

[Jacob Bamberger @jacobbamb]

Flow Matching models often struggle to balance memorization and generalization. 😱
We set out to fix this — by using the geometry of the data manifold.

Introducing Carré du Champ Flow Matching (CDCFM)🧑‍🎨🥖 — improving generalization without sacrificing sample quality.

[Jiajun He @JiajunHe614]

[1/9]🚀Excited to share our new work, RNE! A plug-and-play framework for everything about diffusion model density and control: density estimation, inference-time control & scaling, energy regularisation. More details👇

Joint work with @jmhernandez233 @YuanqiD, Francisco Vargas

[Cristian Bodnar @crisbodnar]

We are looking for three AI Research Engineers and Scientists to join us in our London office at @SilurianAI to build physics foundation models for weather and energy. Apply below and feel free to reach out if you have any questions:

https://t.co/pD2I2aF43m

[Xingyue Huang @hxyscott]

Knowledge Graph Foundation Models (KGFMs) are at the frontier of graph learning - but we didn’t have a principled understanding of what we can (or can’t) do with them. Now we do! 💡🚀
🧵
with Pablo Barcelo, @ismaililkanc, @mmbronstein, @michael_galkin, @JuanLReutter, @OrthMiguel

[Kacper Kapuśniak @KKapusniak1]

We're presenting MFM right now! Come check it out in East Ballroom poster # 2411 🩷🩷🩷 @TeoReu @bose_joey @AlexanderTong7 @mmbronstein @Francesco_dgv

[Teo Reu @TeoReu]

@LogConference @bose_joey @helibenhamu @AlexanderTong7 Graph lovers unite! 🤓✨ Can’t wait to see the generative wizards @bose_joey, @helibenhamu, and @AlexanderTong7 break down Geometric Generative Models like it’s magic. #LoG2024

[Kacper Kapuśniak @KKapusniak1]

@ChurchillMic @PPotaptchik @TeoReu @leoeleoleo1 @AlexanderTong7 @mmbronstein @bose_joey @Francesco_dgv https://t.co/WxlUFtxBTc

[Jacob Bamberger @jacobbamb]

We are happy to announce the Oxford LoG meet-up, LoG-Ox! The event is free to attend and will happen on November 25th.

We plan to have keynote talks, lightning talks, posters and socials. More information and signup form is available here: https://t.co/OlJPxVUTR7.
@LogConference

[Kacper Kapuśniak @KKapusniak1]

I am happy to announce that our paper called Metric Flow Match (MFM) has been accepted to @NeurIPSConf #neurips2024 🔥🔥🔥

I am grateful to all my collaborators and see u in Vancouver! ⛰️

@TeoReu @AlexanderTong7 @mmbronstein @bose_joey @Francesco_dgv

[Teo Reu @TeoReu]

🦕Proving Approximation Results with minimal assumptions in VP-SDE Diffusion. Explore how SDE choices impact sample quality and gain insights into Score Matching in Denoising Diffusion Models! 🦕
Link: https://t.co/R5bzj8Bhyq
Presented at UAI2024!

[Kacper Kapuśniak @KKapusniak1]

If data lives on a manifold, how do we design meaningful interpolations between marginals? We present Metric Flow Matching (MFM)…

@PPotaptchik @TeoReu @leoeleoleo1 @AlexanderTong7 @mmbronstein @bose_joey @Francesco_dgv

🔗Dive in here: https://t.co/XwaU2nyaMT
🧵 (1/12)

[İsmail İlkan Ceylan @ismaililkanc]

Strongly recommended: Emily’s first paper during her PhD. Emily is not on X but she will be in Vienna for #ICML2024!

[Teo Reu @TeoReu]

@mmbronstein are you coming to your birthday party?

[Michael Bronstein @mmbronstein]

A cozy Oxford LoG meetup with guests from London and the Other Place

[Teo Reu @TeoReu]

@hxyscott @ismaililkanc Amazing work done by my lab mate!

[Xingyue Huang @hxyscott]

I am excited to announce that our latest work "A Theory of Link Prediction via Relational Weisfeiler-Leman on Knowledge Graphs", together with Miguel Romero, @ismaililkanc, and Pablo Barceló, will be presented in #NeurIPS 2023.

Arxiv link: https://t.co/3sATH9xu6n

(1/2)🧵