Lightnews — Scholar-powered news

Joan Serrà

@serrjoa.bsky.social

310 followers 150 following 23 posts

Does research on machine learning at Sony AI, Barcelona. Works on audio analysis, synthesis, and retrieval. Likes tennis, music, and wine. https://serrjoa.github.io/

serrjoa.github.io

Posts Media Videos Starter Packs

Pinned

Joan Serrà @serrjoa.bsky.social · Jan 8

I think I may switch back to Twitter/X. Somehow I feel this site didn't take off and I really don't want to be looking at two feeds all the time...

3 3

Joan Serrà @serrjoa.bsky.social · Feb 16

I don't know. I could just now...

Joan Serrà @serrjoa.bsky.social · Jan 8

I think I may switch back to Twitter/X. Somehow I feel this site didn't take off and I really don't want to be looking at two feeds all the time...

3 3

Reposted by Joan Serrà

Dmytro Mishkin @ducha-aiki.bsky.social · Jan 2

Image matching and ChatGPT - new post in the wide baseline stereo blog.

tl;dr: it is good, even feels like human, but not perfect.
ducha-aiki.github.io/wide-baselin...

ChatGPT and Image Matching – Wide baseline stereo meets deep learning

Are we done yet?

ducha-aiki.github.io

2 8 34

Reposted by Joan Serrà

Andrew Gordon Wilson @andrewgwils.bsky.social · Dec 28

Many of the greatest papers, now canonical works, have a story of resistance, tension, and, finally, a crucial advocate. It's shockingly common. Why is there a bias against excellence? And what happens to those papers, those people, when no one has the courage to advocate?

1 2 12

Joan Serrà @serrjoa.bsky.social · Dec 23

Apply here: sonyglobal.wd1.myworkdayjobs.com/Sony_Europe_...

Intern - Machine Learning for Audio

The Music Technology team at Sony AI Barcelona is looking for Research Interns who are passionate about machine learning for audio signal processing. Our mission is to research and develop technologie...

sonyglobal.wd1.myworkdayjobs.com

Joan Serrà @serrjoa.bsky.social · Dec 23

Preferred qualifications:
- PhD candidate or Postdoc.
- Experience with representation/contrastive learning or generative music models.
- Strong programming skills.
- Strong mathematical background.
- Python, github, pytorch, ...
- EU residence permit.
👇

1 1

Joan Serrà @serrjoa.bsky.social · Dec 23

Topics: representation learning for music matching or generative models for music copyright.
Location: Barcelona, on-site (two days a week at least).
Duration: 4-6 months.
Start date: April-November 2025.
Dedication: full-time (part-time also an option).
👇

1 1

Joan Serrà @serrjoa.bsky.social · Dec 23

Do you want to work with me for some months? Two internship positions available at the Music Team of Sony AI in Barcelona!
👇

Views from the office window. Photo taken just now.

1 4 11

Joan Serrà @serrjoa.bsky.social · Dec 21

Haha, me maybe not, but someone should go...

Joan Serrà @serrjoa.bsky.social · Dec 21

Thanks.

Joan Serrà @serrjoa.bsky.social · Dec 21

Congrats to my colleagues, many of whom are not on this website!

Joan Serrà @serrjoa.bsky.social · Dec 21

I'm happy to have two papers accepted at #ICASSP2025!

1) Contrastive learning for audio-video sequences, exploiting the fact that they are *sequences*: arxiv.org/abs/2407.05782

2) Knowledge distillation at *pre-training* time to help generative speech enhancement: arxiv.org/abs/2409.09357

2 15

Joan Serrà @serrjoa.bsky.social · Dec 20

Flow matching mapping text to image directly (instead of noise to image): cross-flow.github.io

Reposted by Joan Serrà

Alexander Kolesnikov @kolesnikov.ch · Dec 20

With some delay, JetFormer's *prequel* paper is finally out on arXiv: a radically simple ViT-based normalizing flow (NF) model that achieves SOTA results in its class.

Jet is one of the key components of JetFormer, deserving a standalone report. Let's unpack: 🧵⬇️

2 7 42

Reposted by Joan Serrà

Deep Learning Barcelona @dlbcnai.bsky.social · Dec 19

Did you miss any of the talks of the Deep Learning Barcelona Symposyum 2024 ? Play them now from the recorded stream:

www.youtube.com/live/yPc-Un3...

YouTube

Share your videos with friends, family, and the world

www.youtube.com

1 4

Reposted by Joan Serrà

Jeremy Howard @howard.fm · Dec 19

I'll get straight to the point.

We trained 2 new models. Like BERT, but modern. ModernBERT.

Not some hypey GenAI thing, but a proper workhorse model, for retrieval, classification, etc. Real practical stuff.

It's much faster, more accurate, longer context, and more useful. 🧵

19 150 620

Joan Serrà @serrjoa.bsky.social · Dec 19

On pre-acrivation norm, learnable residuals, etc.

Lucas Beyer (bl16) @giffmana.ai · Dec 18

A post by @cloneofsimo on Twitter made me write up some lore about residuals, ResNets, and Transformers. And I couldn't resist sliding in the usual cautionary tale about small/mid-scale != large-scale.

Blogpost: lb.eyer.be/s/residuals....

Reposted by Joan Serrà

Kyle Kastner @kastnerkyle.bsky.social · Dec 18

Two great tokenizer blog posts that helped me over the years: sjmielke.com/papers/token...

sjmielke.com/comparing-pe...

People have mostly standardized on certain tokenizations right now, but there are huge performance gaps between locales with high agglomeration (e.g. common en-us) and ...

1 3 10

Joan Serrà @serrjoa.bsky.social · Dec 16

No.

Joan Serrà @serrjoa.bsky.social · Dec 15

Don't be like Reviewer 2.

Reposted by Joan Serrà

Peyman Milanfar @docmilanfar.bsky.social · Dec 14

Did Gauss invent the Gaussian?

- Laplace wrote down the integral first in 1783
- Gauss then described it in 1809 in the context of least-sq. for astronomical measurements
- Pearson & Fisher framed it as ‘normal’ density only in 1910

* Best part is: Gauss gave Laplace credit!

5 38

Joan Serrà @serrjoa.bsky.social · Dec 13

I already signed up (as a mentor) for this year!

Deep Learning Barcelona @dlbcnai.bsky.social · Dec 13

Call for mentees and mentors open until December 16.

Sign up as a mentee if you are a student or in the early stages of your career.

Sign up as a mentor to help in the career growth of a member of the #DLBCN community.

Details and registration:
sites.google.com/view/dlbcn20...

Reposted by Joan Serrà

Jörg Franke @jfranke.bsky.social · Dec 9

Thrilled to present our work on Constrained Parameter Regularization (CPR) at #NeurIPS2024!
Our novel deep learning regularization outperforms weight decay across various tasks. neurips.cc/virtual/2024...
This is joint work with Michael Hefenbrock, Gregor Köhler, and Frank Hutter
🧵👇

NeurIPS Poster Improving Deep Learning Optimization through Constrained Parameter RegularizationNeurIPS 2024

neurips.cc

1 1 2

Reposted by Joan Serrà

Keenan Crane @keenancrane.bsky.social · Dec 9

Entropy is one of those formulas that many of us learn, swallow whole, and even use regularly without really understanding.

(E.g., where does that “log” come from? Are there other possible formulas?)

Yet there's an intuitive & almost inevitable way to arrive at this expression.

22 130 550

Reposted by Joan Serrà

Tanishq Mathew Abraham @iscienceluvr.bsky.social · Dec 10

Inventors of flow matching have released a comprehensive guide going over the math & code of flow matching!

Also covers variants like non-Euclidean & discrete flow matching.

A PyTorch library is also released with this guide!

This looks like a very good read! 🔥

arxiv: arxiv.org/abs/2412.06264

1 26 110