On Diffusion Posterior Sampling via Sequential Monte Carlo for Zero-Shot Scaffolding of Protein Motifs • This study introduces a novel framework for zero-shot protein motif scaffolding using diffusion posterior sampling with Sequential Monte Carlo (SMC), leveraging the Genie

Imagine my surprise at learning that the Online Safety Act is why OpenAI has blocked Sora in the UK, putting us in the same impoverished position as the European Union. ofcom.org.uk/online-safety/…

Bank CEOs should be forced to have a framed version of this image on their desks.

Gaussian entropic optimal transport: Schr\"odinger bridges and the Sinkhorn algorithm ift.tt/qWbU5Oh

Everybody gangsta until SDEs work! #ICLR2025 Noise hits every optimizer differently! For SignSGD, adaptivity increases resistance to gradient noise, while AdamW enjoys extreme stability. Plus, AdamW has an exciting new scaling rule! More below👇! arxiv.org/abs/2411.15958

Non-asymptotic Analysis of Diffusion Annealed Langevin Monte Carlo for Generative Modelling ift.tt/epDhJMc

Very happy to announce the 2nd RSS/Turing Workshop on Gradient Flows for Sampling, Inference, and Learning to take place on 24th of March, 2025! Registration is open now: tinyurl.com/rss-turing-gfl… Please see the webpage for topics and speakers.

My latest tweet barely hit 130 views... Giving it another shot. Is it even worth posting on this platform anymore without a blue tick? x.com/odakyildiz/sta…

O. Deniz Akyildiz, Pierre del Moral, Joaquin Miguez. [math.PR]. On the contraction properties of Sinkhorn semigroups. arxiv.org/abs/2503.09887

Pierre Del Moral in action: arxiv.org/abs/2503.15963 What a beast.

Training Latent Diffusion Models with Interacting Particle Algorithms. arxiv.org/abs/2505.12412

Video from the 2nd RSS/Turing workshop on gradient flows for sampling, inference, and learning: youtu.be/QrPy49eAe9o?fe…

James Cuin, Davide Carbone, O. Deniz Akyildiz. [stat.CO]. Learning Latent Variable Models via Jarzynski-adjusted Langevin Algorithm. arxiv.org/abs/2505.18427

An unexpected consequence of mass llm use: fake citations of me show up in Google scholar that seem hallucinated from a prompt like "write me a related work". Showing up means the SAME hallucination was published twice! I find this hilarious but this seems a bad sign for academia

Proud advisor moment! Find Paula Cordero's paper here: openreview.net/forum?id=rTqyD… x.com/UncertaintyInA…

congratulations to Paula Cordero Encinar, Francesca Romana Crucinio and O. Deniz Akyildiz (O. Deniz Akyildiz) for winning the 🏆 Best Student Paper Award 🏆 at #UAI2025 with "Proximal Interacting Particle Langevin Algorithms" 👉 openreview.net/forum?id=rTqyD…

Paula Cordero-Encinar, Andrew B. Duncan, Sebastian Reich, O. Deniz Akyildiz. [stat.CO]. Sampling by averaging: A multiscale approach to score estimation. arxiv.org/abs/2508.15069

A Gradient Flow Approach to Solving Inverse Problems with Latent Diffusion Models ift.tt/F0eYzXl

Uniform-in-time convergence bounds for Persistent Contrastive Divergence Algorithms. arxiv.org/abs/2510.01944

Learning Latent Energy-Based Models via Interacting Particle Langevin Dynamics ift.tt/L7Stpes