Stathi Fotiadis

Machine Learning for Physical Sciences @imperialcollege

[Stathi Fotiadis @stathiuss]

@bcherny Does this work with yolo mode?

[Stathi Fotiadis @stathiuss]

Check out our new paper on super resolution of misaligned data.

[Morteza Mardani @MardaniMorteza]

🔍 Superresolution misalignment? Conditional diffusion/flow models falter when low-res and high-res data don’t align.
Meet Stochastic Flow Matching—a framework mapping low-res to latent space, bridging high-res targets effectively.

📝 Paper: https://t.co/LOdfYgpDau

[Stathi Fotiadis @stathiuss]

I have been using sioyek as a pdf reader because it allows one click reference viewing and google scholar search https://t.co/feoNhypSig, but the new Google Scholar PDF reader plugin seems ever better. https://t.co/S343epx6bf

[Stathi Fotiadis @stathiuss]

@BlackHC @clured Any examples of where this helps?

[Stathi Fotiadis @stathiuss]

@wgrathwohl @TPAIN Damn that sounds unbelievable!

[Stathi Fotiadis @stathiuss]

Our new paper on diffusion models improves sampling at very low NFEs.
Accepted @NeurIPSConf Diffusion Workshop. Code coming soon. Props to @guoxxoug

[Ayan Das @dasayan05]

Our paper, accepted at @NeurIPSConf '23 Diffusion Model Workshop, shows a normalisation trick for DEIS, a SoTA diffusion sampler.

Work done primarily by Guoxuan (@guoxxoug) during his internship at @MediaTek Research UK.

🧵1/3

[Stathi Fotiadis @stathiuss]

Anyone else using DelveGPT?

[Stathi Fotiadis @stathiuss]

@Eric_Wallace_ Do you think in unconditional generation the memorization would be higher? How would you go about finding the noise that leads to an image from the train dataset?

[Stathi Fotiadis @stathiuss]

Seems like the limits for the Topology workshop were not well-defined. #icml #ICML2023

[Sander Dieleman @sedielem]

Room 314, join us!

[Sander Dieleman @sedielem]

📢 diffusion circle! 📢

As is becoming tradition, let's talk diffusion/iterative refinement at #ICML2023!

Let's meet at registration 3:00pm on Thursday July 27 and find a spot to sit and chat. Please share and tag anyone at the conference who might be interested!

[Stathi Fotiadis @stathiuss]

@LinoMarioA Thanks for being such a great collaborator!

[Stathi Fotiadis @stathiuss]

#ICML2023 #ICML

[Stathi Fotiadis @stathiuss]

Attending @icmlconf and will present two posters tomorrow:

1) Image generation with shortest path diffusion
at 11am with @dasayan05, paper: https://t.co/J80kXcvhlI

2) Disentangled generative models for robust dynamical system prediction
at 2pm, paper: https://t.co/Z2ts5yQ5vl

[Stathi Fotiadis @stathiuss]

Glad we established that. #icml #ICML2023

[Stathi Fotiadis @stathiuss]

@sstj389 @rnnishard Glad to hear you find our work interesting! The code can be found here: https://t.co/RZq3oC5GF9

[Stathi Fotiadis @stathiuss]

@MilesCranmer @andrewwhite01 @jobovy @JayWadekar1 Pretty neat. Does it parse pdf correctly?

[Stathi Fotiadis @stathiuss]

@y0b1byte The contents page resonates well with the experience. Can't wait to read the rest.

[Mario Lino @LinoMarioA]

Our last paper published by @AIP_Publishing introduces a rotation-equivariant multi-scale graph neural network. This data-driven model can be used to simulate fluid dynamics in real-time and to speed up design-space exploration.
https://t.co/Gl12EbflKR

[arXiv Daily @Arxiv_Daily]

Towards Fast Simulation of Environmental Fluid Mechanics with Multi-Scale Graph Neural Networks
https://t.co/5HMn6Wz0Bo
by @LinoMarioA et al. including @stathiuss
#MachineLearning #NeuralNetwork