Danilo J. Rezende @danilojrezende.bsky.social · 27/12/2024Indeed, additionally If computing the forces in HMC is computationally cheap - it's extremely hard to beat it with learnt samplers in terms of wall clock time. Similar challenge to Neural PDE solvers. 040
Reposted by Danilo J. RezendeGuillem Simeon @guillemsimeon.bsky.social · 25/12/2024Happy Newton’s birthday to those who celebrate 1111
Reposted by Danilo J. RezendeSander Dieleman @sedielem.bsky.social · 02/12/2024In arxiv.org/abs/2303.00848, @dpkingma.bsky.social and @ruiqigao.bsky.social had suggested that noise augmentation could be used to make other likelihood-based models optimise perceptually weighted losses, like diffusion models do. So cool to see this working well in practice!arxiv.orgUnderstanding Diffusion Objectives as the ELBO with Simple Data AugmentationTo achieve the highest perceptual quality, state-of-the-art diffusion models are optimized with objectives that typically look very different from the maximum likelihood and the Evidence Lower Bound (... 05311
Reposted by Danilo J. RezendeAndrew Gordon Wilson @andrewgwils.bsky.social · 18/12/2024This is your monthly reminder that understanding deep learning does not require rethinking generalization, and it never did. 3223
Reposted by Danilo J. RezendeDhruv Batra @dhruvbatra.bsky.social · 14/12/2024 Brilliant talk by Ilya, but he's wrong on one point. We are NOT running out of data. We are running out of human-written text. We have more videos than we know what to do with. We just haven't solved pre-training in vision. Just go out and sense the world. Data is easy. 59916
Danilo J. Rezende @danilojrezende.bsky.social · 07/12/2024More a linear algebra meme than a QM meme 😅 141
Reposted by Danilo J. RezendeFrank Noe @franknoe.bsky.social · 06/12/2024Super excited to preprint our work on developing a Biomolecular Emulator (BioEmu): Scalable emulation of protein equilibrium ensembles with generative deep learning from @msftresearch.bsky.social ch AI for Science. www.biorxiv.org/content/10.1... 21441147
Reposted by Danilo J. RezendeKyle Cranmer @kylecranmer.bsky.social · 02/12/2024My former student @msalbergo.bsky.social has beautiful papers describing the approach on the right. “Flow matching” and “stochastic interpolants” were concurrent developments of the same core idea. arxiv.org/abs/2209.15571 arxiv.org/abs/2303.08797 2404
Reposted by Danilo J. RezendeMiles Cranmer @milescranmer.bsky.social · 02/12/2024🧵 Today with @polymathicai.bsky.social and others we're releasing two massive datasets that span dozens of fields - from bacterial growth to supernova! We want this to enable multi-disciplinary foundation model research. 38825
Danilo J. Rezende @danilojrezende.bsky.social · 30/11/2024As a statistical inequality hoarder, I approve 😅 030
Reposted by Danilo J. RezendeGuillaume Dalle @gdalle.bsky.social · 30/11/2024A paper a day, episode 15. You liked the matrix cookbook? You’re gonna love this one. 100 statistics inequalities just for your personal enjoyment. As they say in French, moi j’ai Bienaymé cet article ! arxiv.org/abs/2102.07234arxiv.orgOne Hundred Probability and Statistics InequalitiesHerein we present one hundred inequalities culled from various corners of the probability, statistics, and combinatorics literature. We welcome new suggestions. 55910
Danilo J. Rezende @danilojrezende.bsky.social · 30/11/2024What do you mean by physical? Eg If the target density lives on a specific manifold, RFs will generally not respect that naively (eg think of data living on a torus). One would have to choose paths along geodesics for example and then properly integrate the flow on the manifold. 120
Danilo J. Rezende @danilojrezende.bsky.social · 30/11/2024Yes, though a simplification is that the log-det-jac is the integral of the divergence. So "just" need to solve a 1d integral. 010
Danilo J. Rezende @danilojrezende.bsky.social · 30/11/2024You can, just need to integrate it's divergence to get the likelihood. 010
Danilo J. Rezende @danilojrezende.bsky.social · 29/11/2024What do you mean? That's the whole field of optimisation (preconditioning, proximal, natural, etc...) 040
Reposted by Danilo J. RezendeGuillaume Dalle @gdalle.bsky.social · 26/11/2024For the theoretical side, there’s no better resource than the 2024 book by Blondel and Roulet: arxiv.org/abs/2403.14606 For the practical side, my friend @willtebbutt.bsky.social wrote stellar documentation explaining autodiff with mutation: compintell.github.io/Mooncake.jl/...arxiv.orgThe Elements of Differentiable ProgrammingArtificial intelligence has recently experienced remarkable advances, fueled by large models, vast datasets, accelerated hardware, and, last but not least, the transformative power of differentiable p... 55612
Danilo J. Rezende @danilojrezende.bsky.social · 25/11/2024Not sure it works well here neither. Still seeing lots of cartoons and cat pictures instead of papers. 210
Reposted by Danilo J. RezendeSimon Batzner @simonbatzner.bsky.social · 25/11/2024We're hiring! Our team at Google DeepMind is looking for a research engineer to join us. More details in the link below. boards.greenhouse.io/deepmind/job...boards.greenhouse.ioResearch Engineer, Materials ScienceLondon, UK 0346
Danilo J. Rezende @danilojrezende.bsky.social · 25/11/2024A cargo train would probably be the most efficient by far? 000
Reposted by Danilo J. RezendeSurya Ganguli @suryaganguli.bsky.social · 25/11/2024This video takes pedagogy to a whole new level in teaching the Schrödinger equation! youtu.be/uVKMY-WTrVo?...youtu.beWhat is the i really doing in Schrödinger's equation?YouTube video by Welch Labs 47713
Reposted by Danilo J. RezendeJennifer Johnson @jajohnson.bsky.social · 18/11/2024Updated periodic table with origin of elements in the Solar System (for boron and lithium at least) 45950294