Adrián Javaloy

Adrián Javaloy

@javaloyml.bsky.social

Lecturer at the University of Bath working on Machine Learning. Previously in Edinburgh, Saarbrücken, Tübingen, and Murcia. 🌐 adrianjav.github.io

1.6K followers424 following94 posts

Adrián Javaloy reposted

Funding opportunities for a fully-funded PhD at the University of Bath (UK) for Chinese Nationals. Please get in touch if you would like to discuss a PhD in video understanding and explainable computer vision. Info here: www.bath.ac.uk/campaigns/fu...

bath.ac.ukFunding opportunities for new doctoral students from ChinaFind out about the funding available to help you fund your postgraduate research and study as a new doctoral student from China.

More importantly, we show that this comes at no cost in performance, and we can even train non-structured-decomposable squared circuits* (* That is, circuits which cannot be efficiently squared)

Yes, we can! 💡 We generalize both ideas and propose to use orthogonality constraints to parametrize *already normalized* squared circuits That way, we completely avoid squaring them during training!

As a result, we can train really large squared circuits while saving both time and memory! At 357M parameters, we (⟂) use: - 12 GiB vs 18 GiB (33% reduction!) - 0.29ms vs 0.52ms per iteration (44% faster!)

One way of increasing the expressiveness of probabilistic circuits is to square them (multiply a circuit with itself). 😔 However, this imposes a quadratic cost in the circuit size, as we need to re-normalize it to ensure that it encodes a valid probability.

In the tensor network community, a similar issue can be avoided for specific cases using canonical forms And in the circuit community, determinism (i.e. non-overlapping supports) makes the square tractable, although it is too restrictive... 🤔 Can we go expand on these ideas?

Adrián Javaloy reposted

Want to use your favourite #NeSy model but afraid of the reasoning shortcuts?🫣 Fear not💪🏻In our #NeurIPS2025 paper we show that you just need to equip your favourite NeSy model with prototypical networks and the reasoning shortcuts will be a problem of the past!

Adrián Javaloy reposted

📈 Tenerife Norte bate su récord de temperatura para noviembre: 33 °C el día 4. ➡️ Supera ampliamente el registro máximo anterior, de 31 °C. Tenerife Norte tiene una serie de datos de 85 años.

Adrián Javaloy reposted

Does a smaller latent space lead to worse generation in latent diffusion models? Not necessarily! We show that LDMs are extremely robust to a wide range of compression rates (10-1000x) in the context of physics emulation. We got lost in latent space. Join us 👇

View full thread

Also, I did put considerably effort to frame everything under a common question: > What biases can we add to DL optimization so that the outcome of the model is what we expected from the beginning? 🧵2/4

What a nice experience! Thank you everyone who attended TPM! Particularly those who engaged in the poster sessions, few times I had so much fun discussing my poster!

uai2026@auai.org

memories from the last day of #UAI2025: the 8th Workshop on Tractable Probabilistic Modeling #TPM2025 tractable-probabilistic-modeling.github.io/tpm2025/

Adrián Javaloy reposted

likely one of the best editions of #TPM ever! big thanks to @poorvagarg.bsky.social @jsleland.bsky.social @javaloyml.bsky.social @zzhe.bsky.social @lennertds.bsky.social Lingyun Yao and Christoph Staudt for organizing it and to everyone who attended it!

uai2026@auai.org

memories from the last day of #UAI2025: the 8th Workshop on Tractable Probabilistic Modeling #TPM2025 tractable-probabilistic-modeling.github.io/tpm2025/

Next week I will be in 🌴 Rio 🌴 attending (and helping organizing) UAI 2025 and TPM 2025. If you are also attending and want to talk science (e.g. probabilistic generative models, causality, multi-objective machine learning, etc) or just hang out, not everything is work, feel free to reach me out!

Adrián Javaloy@javaloyml.bsky.social

This year I am co-organizing the 8th iteration of the Tractable Probabilistic Modeling #TPM workshop at #UAI2025 🌴 Rio de Janeiro edition 🌴 🌐 lnkd.in/dDK8T5Au ⏰ Submission deadline: May 23th AoE 🌴 Date: July 15th 🧵👇

Unrelatedly, the rule of mandatory physical attendance for authors was changed *after* the submission deadline: web.archive.org/web/20250514... People submitted their works under some given premises, altering them a posteriori is questionable at best...

NeurIPS Conference@neuripsconf.bsky.social

NeurIPS is endorsing EurIPS, an independently-organized meeting which will offer researchers an opportunity to additionally present NeurIPS work in Europe concurrently with NeurIPS. Read more in our blog post and on the EurIPS website: blog.neurips.cc/2025/07/16/n... eurips.cc

Adrián Javaloy reposted

✨Thrilled to see EurIPS launch — the first officially endorsed European NeurIPS presentation venue! 👀 But NeurIPS now requires at least one author to attend in San Diego or Mexico (and not just virtually as before). This is detrimental to many. Why not allow presenting at EurIPS or online? 1/4

NeurIPS Europe@neuripseurope.bsky.social

EurIPS is coming! 📣 Mark your calendar for Dec. 2-7, 2025 in Copenhagen 📅 EurIPS is a community-organized conference where you can present accepted NeurIPS 2025 papers, endorsed by @neuripsconf.bsky.social and @nordicair.bsky.social and is co-developed by @ellis.eu eurips.cc

Adrián Javaloy reposted

For years, people have suggested having alternate NeurIPS locations, outside the US. 2025 is the first time it will be done, with satellite locations in Copenhagen and Mexico City. I wonder if this will be a one-off thing, or whether the trend will continue.

Really great news! 👏

NeurIPS Europe@neuripseurope.bsky.social

EurIPS is coming! 📣 Mark your calendar for Dec. 2-7, 2025 in Copenhagen 📅 EurIPS is a community-organized conference where you can present accepted NeurIPS 2025 papers, endorsed by @neuripsconf.bsky.social and @nordicair.bsky.social and is co-developed by @ellis.eu eurips.cc

Adrián Javaloy reposted

Adrián Javaloy reposted

❗📅 Extended deadline: May 29th 👥 Several positions: - Postdoc ( Project 1: AI for Earth System Modeling) - PhD, Postdoc, Scientific Programmer (Project 2: Next-Gen Foundation Models) - PhD + Postdoc (Project 3 – AI for Scientific Discovery) 📝 Apply here: forms.gle/grkEhmXrDout...

docs.google.comdocs.google.com

🚀 AI & Earth Science Research Positions – Apply Now! 🌍🤖 Join us at the University of Valencia for #PhD, #postdoc & scientific programmer roles in AI-driven Earth sciences! 📅 Deadline: April 15, 2025 🔗 forms.gle/jg9Z181yQKrZ... #AI #Causality #discovery #EarthSciences

We have extended the submission deadline for TPM to the 2nd of June! Remember that **we accept recently accepted works that fit in topic,** so if you want to travel to Rio and share your work it's a no brainer imho 🌴🌴

Adrián Javaloy@javaloyml.bsky.social

This year I am co-organizing the 8th iteration of the Tractable Probabilistic Modeling #TPM workshop at #UAI2025 🌴 Rio de Janeiro edition 🌴 🌐 lnkd.in/dDK8T5Au ⏰ Submission deadline: May 23th AoE 🌴 Date: July 15th 🧵👇

Adrián Javaloy reposted

📢 I am looking for AI post-doc/research/engineer positions in Europe (Paris, London, Zurich, ...) starting 2026. My work revolves around generative modeling and AI for Science, with 4+ publications at top conferences during my PhD. If you are hiring, please reach out! If not, please repost 🔁

Adrián Javaloy reposted

Join us for our next I-X seminar with Dr Antonio Vergari @nolovedeeplearning.bsky.social titled "Logically-Consistent Deep Learning". 🕓 13:00 - 14:00 (GMT) 📅 Thursday, 5 June 📍 Hybrid (White City Campus / Microsoft Teams) Register via link below. www.imperial.ac.uk/events/19426...

imperial.ac.ukI-X Seminar: Logically-Consistent Deep Learning with Dr Antonio VergariIn this talk, Dr Antonio Vergari (University of Edinburgh) will speak about logically-consistent deep learning.

Adrián Javaloy reposted

Rio is looking forward to seeing all your works on #fast #reliable #tractable #models #circuits #tensor #networks #flows! please share!

antonio vergari ⚔️ short-circuiting@nolovedeeplearning.bsky.social

the #TPM ⚡Tractable Probabilistic Modeling ⚡Workshop is back at @auai.org #UAI2025! Submit your works on: - fast and #reliable inference - #circuits and #tensor #networks - normalizing #flows - scaling #NeSy #AI ...& more! 🕓 deadline: 23/05/25 👉 tractable-probabilistic-modeling.github.io/tpm2025/

Adrián Javaloy reposted

We propose Neurosymbolic Diffusion Models! We find diffusion is especially compelling for neurosymbolic approaches, combining powerful multimodal understanding with symbolic reasoning 🚀 Read more 👇

Adrián Javaloy reposted