Math+ML+X_Lab

PI: @aviles_rivero. Yau Mathematical Sciences Centre, Tsinghua University.
We focuse on the intersection of Math, ML, and real-world applications (X).

[Math+ML+X_Lab @MATH_ML_X_Lab]

🚀 Exciting news! 🚀

We're launching a brand-new channel! 🎉
Join the fun and stay updated – just scan the QR below to get started! 📲👇

[Math+ML+X_Lab @MATH_ML_X_Lab]

Do deeper neural operators forget geometry?

Our ICML 2026 paper introduces the Geometry Forgetting Hypothesis — a hidden reason why neural operators fail on irregular domains.

Shoutout to PhD student Yanming Xia and our PI for this work 👏

▶️https://t.co/6JYTGNLch7 #ICML2026

[Machine Learning (ML) Papers @Memoirs]

Do Neural Operators Forget Geometry? The Forgetting Hypothesis in Deep Operator Learning

Yanming Xia, Angelica I. Aviles-Rivero
https://t.co/GHj2QwFhEH [𝚌𝚜.𝙻𝙶]

[Math+ML+X_Lab @MATH_ML_X_Lab]

🚀 The Math+ML+X Seminar Series is back!
Don’t miss our next talk by Han Zhang on physics-informed AI for blood flow imaging 🧠💡

📅 May 14 | ⏰ 4PM
💻 Voov: 294-173-543

Where math meets machine learning meets real-world impact.
#MathMLX #PINNs #AIforScience

[Math+ML+X_Lab @MATH_ML_X_Lab]

🚨 Today is the Day! ICLR 2026 Oral Presentation
Sharing our work on rethinking attention for medical time series today 👇

🎤 Oral Session 5F — AI for Science II
📍 Room 201C

Would love to chat!!
#ICLR2026

[Math+ML+X_Lab @MATH_ML_X_Lab]

🧵ICLR 2026 Oral: Transformers break on medical time series (e.g., EEG/ECG). Why? Attention assumes all channels interact equally —biology doesn't.

CoTAR uses one shared core token.
✅ 5x faster | 70% less memory | +10% accuracyMore connections ≠ better!

#ICLR2026

[Math+ML+X_Lab @MATH_ML_X_Lab]

🧵ICLR 2026 Oral: Transformers break on medical time series (e.g., EEG/ECG). Why? Attention assumes all channels interact equally —biology doesn't.

CoTAR uses one shared core token.
✅ 5x faster | 70% less memory | +10% accuracyMore connections ≠ better!

#ICLR2026

[Math+ML+X_Lab @MATH_ML_X_Lab]

Proud moment 🎉 Our PI at XInfinity, sharing how math + ML reveal “infinite possibilities” hidden in data—core to our Math+ML+X Lab @ YMSC, Tsinghua!

[Angelica I Aviles-Rivero @aviles_rivero]

The stage is set 🔥
ISBI 2026 starts tomorrow—who’s ready?
Come say hi and let’s connect!
@IeeeIsbi

[Math+ML+X_Lab @MATH_ML_X_Lab]

���Are you attending ISBI 2026? @IeeeIsbi 👀
The Math+ML+X Lab is warming up (yes, gradients included).
Come find us, come say hi, exchange ideas, and connect
Bonus: we’re hiring — so let’s chat!

[Math+ML+X_Lab @MATH_ML_X_Lab]

Happy #PiDay!

A wild fact:
Pick two random integers.
The probability they share no common divisor is
6 / π² ≈ 0.6079
A constant discovered while studying circles suddenly appears in number theory!

π also shows up in machine learning!
#InternationalDayofMathematics

[Math+ML+X_Lab @MATH_ML_X_Lab]

Operator learning + geometry + transformers = 🔥

Excited to host Shizheng Wen (ETH Zürich) at the Math+ML+X Seminar next Friday !

If PDEs and ML are your thing, don’t miss it.

🗓 March 20 | ⏰ 4pm | 💻 Online

[Math+ML+X_Lab @MATH_ML_X_Lab]

Happy International Women’s Day! 🌸

Math+ML+X Lab celebrates the strength, creativity, and resilience of women everywhere.

We honour the countless ways women contribute to our communities, inspire future generations, and help shape a better world!!
#InternationalWomensDay

[MICCAI Students @MiccaiStudents]

Calling students & early-career researchers!
Learn how to turn ideas & experiments into a strong #MICCAI2026 submission at the MSB Webinar panel discussion.
📅 Jan 21, 8:00 PM EST | Jan 22, 2:00 AM CET / 9:00 AM CST
🔗https://t.co/xaCcfSjTtK

[ELLIS @ELLISforEurope]

🔄 Revisit the profiles of 50 Women in ELLIS to discover women leading AI research across Europe—their achievements, impact and advice.

👉 https://t.co/4dMXGPHkEy

[Math+ML+X_Lab @MATH_ML_X_Lab]

Into distributions, slices, and reconstruction algorithms?
We’ve got something you’ll want to see at Poster #1806 at #NeurIPS2025.
📅 Dec 5 | ⏰ 11–2 | 📍 Exhibit Hall C/D/E

[Math+ML+X_Lab @MATH_ML_X_Lab]

🎴Big news! Our FNO-based approach accelerates quantum-fluid modelling by ~4000× — while preserving full physical accuracy.
Now published in Nature Portfolio’s Communications Physics!

Huge thanks to our incredible collaborators who made this possible!!
https://t.co/r7cA3Ynkpw

[Math+ML+X_Lab @MATH_ML_X_Lab]

📢 News! (cc: @MICCAI_Society @IeeeIsbi) @aviles_rivero & @WangShujunCUHK will co-lead Winter School 2025: Medical Imaging in the Deep Learning Era!

✨ Two days of AI + medical imaging
✨ Amazing speaker lineup
✨ FREE & online

🗓️ Dec 5–6, 2025
🔗 https://t.co/S2YXRsV3Qr

[Math+ML+X_Lab @MATH_ML_X_Lab]

🚨 NeurIPS packing mode: ON. Bringing D2SA — come say hi!

Dual-Stage MRI TTA:
🔭 Global implicit representation for domain shift
🧩 Diffusion-based refinement for crisp anatomy

✨ Operator-shift-aware
✨No TTA collapse
✨Strong OOD robustness

Team crushed it! 🙌 #neurips2025

[Math+ML+X_Lab @MATH_ML_X_Lab]

🔊 We’re hiring Interns / RAs / Postdocs to solve for x:

x ∈ {Inverse Problems, Computer Vision, Learned PDEs, Graph Learning, LLMs4Science, AI4Science, ...}
s.t. x ↦ real impact {healthcare, weather, materials, + more! } 🌎

🏆Join before the gradient vanishes. ∇💥 !!

[Math+ML+X_Lab @MATH_ML_X_Lab]

🚨 Schedule Update!
Our MATH+ML+X Seminar with Prof. Hongxia Yang from The Hong Kong Polytechnic University starts in less than 1 hour!!! ⏰

🧠 Topic: Co-GenAI: A Novel Fusion-Driven Framework
🗓️ Today at 4 PM (Beijing Time)
💻 Join via Voov (tencent): 177-832-355

[Math+ML+X_Lab @MATH_ML_X_Lab]

🚀 Our Math+ML+X group at #MICCAI2025:

🏆 PI won the Outstanding Area Chair Award!, served on Best Paper committee & joins the #MICCAI2026 Committee!
📚 We co-organised the Graph Learning Tutorial & Workshop on MedAGI
🧮 Presented Implicit U-KAN2.0 for medical segmentation!