Anirban Ray

Exciting news! An article on India BioImaging (IBI) has been published in "Microscopy and Analysis" (@wileylifesci.bsky.social )(Issue 2/2026)

Featured under "Building the India Bioimaging Community," it highlights IBI's vision, activities & growing impact in strengthening bioimaging across India.

Congratulations 🎊🍾🎉

Excited to share our new paper (CVPR 2026 🚀): "MuViT: Multi-Resolution Vision Transformers for Learning Across Scales in Microscopy" which enables local predictions to use global context.
Great work led by @albertdm.bsky.social & another fun collab w @gioelelamanno.bsky.social! @scadsai.bsky.social

It was super cool to hear @alex-krull.bsky.social talking about his research at @humantechnopole.bsky.social. Alex never fails to inspire his audience 🫠🤩

yeah exactly! I think that the training objective (minimizing drift between p and q) will inherently try to prioritize perception more than distortion (low FID), and with one-shot prediction, I feel that the chance of hallucination looks high (similar to GANs), but let's see!

haha, true! Aside from these, I find the idea of training time distribution evolution intriguing. Still trying to understand the implication of drifting models for inverse problems tho :)

Excellent talk by Matteo Hessel from Google DeepMind at @humantechnopole.bsky.social earlier today!
Thanks for dropping by and infecting us with the Meta-RL bug!

🤩

#isbi2026 #generativeai #flowmatching #computationalsuperresolution #bioimageanalysis | Anirban Ray Excited to share that ResMatching (https://lnkd.in/dn_6VB9Q) has been accepted to IEEE International Symposium on Biomedical Imaging (ISBI) 2026! 🎉🎉 This work represents a significant step forward in...

ResMatching accepted to IEEE ISBI 2025🎉. We extend Guided CFM to Computational Super-Resolution under extreme noise, achieving SOTA performance and calibrated posterior sampling that reflects uncertainty in real data. Learn more here: www.linkedin.com/posts/anirba...

YouTube video by Jia-Bin Huang How Residual Connections Are Getting an Upgrade [mHC]

Residual connections are adopted in virtually every deep learning model.

BUT, can we further improve it? Hyper-connections is an exciting recent exploration to generalize residual connections.

Check out the video explaining manifold constraints hyper-connections!
youtu.be/jYn_1PpRzxI

Self-Supervised Learning from Noisy and Incomplete Data Many important problems in science and engineering involve inferring a signal from noisy and/or incomplete observations, where the observation process is known. Historically, this problem has been tac...

📖 We put together with Mike Davies a review of self-supervised learning for inverse problems, covering the main approaches in the literature with a unified notation and analysis.

arxiv.org/abs/2601.03244

Image of the Human Technopole https://it.wikipedia.org/wiki/Human_Technopole#/media/File:Human_Technopole_Milano.jpg

I'm super excited to announce that I will join the Human Technopole @humantechnopole.bsky.social in Milan in April 2026 as a Research Group Leader.

Deep Learning for Microscopy Image Analysis Topics The following will be covered extensively during lectures, exercises, and project work: Image denoising and restoration (fully supervised and self-supervised) Image translation (e.g.,

🚨 Alarm!!! 🚨
AI/ML course for microscopy image analysis!!! 🧐
In 2026 at Janelia (@hhmijanelia.bsky.social), no tuition, housing and meals provided! Isn’t that borderline unbelievable?!?

20 students, ~14 TAs and lecturers
🗓️ June 4-18 2026

✍️ Jan 15 2026 ✍️

🔁 pls!!

www.janelia.org/you-janelia/...

Welcome to @humantechnopole.bsky.social, Jan 🤩

Okay, maybe I didn't pay much attention to this before but there are some elegant works that truly amaze me. This is one of them. A task-agnostic setup built on flow matching that just works across problems. I really like this work. 😀😃

Invited talk by Rich Sutton at @neuripsconf.bsky.social was presenting the OaK architecture.
A “scaffold” for superintelligent agents in which human inputs (at design time) are depreciated. 🧵

"Back to Basics: Let Denoising Generative Models Denoise" by
Tianhong Li & Kaiming He arxiv.org/abs/2511.13720
Diffusion models in pixel-space, without VAE, with clean image prediction = nice generation results. Not a new framework but a nice exploration of the design space of the diffusion models.

We moved the AI@MBL course "Deep Learning for Microscopy Image Analysis" to HHMI Janelia (@hhmijanelia.bsky.social).

Join us for two weeks of intense lectures, exercise, and hands-on project work!

Course dates: June 4-18 2026
Application by: January 15 2026

www.janelia.org/you-janelia/...

Congratulations 🎉🎊

💥 DeepInverse is now part of the official PyTorch Landscape💥

We are excited to join an ecosystem of great open-source AI libraries, including @hf.co diffusers, MONAI, einops, etc.

pytorch.org/blog/deepinv...

I’m at #AIS25 today and tomorrow…
Where scientists, industry leaders, investors, and policymakers meet to explore the transformative impact of artificial intelligence on scientific discovery.
I think this is a very important conversation we must have NOW! 👍

Anirban Ray, Vera Galinova, Florian Jug
ResMatching: Noise-Resilient Computational Super-Resolution via Guided Conditional Flow Matching
https://arxiv.org/abs/2510.26601

needless to say! and much appreciated 😊

super cool work😎

“We may not win every battle, but we will win the war.” --- Such an appropriate characterization for posterior samplers. Each posterior sample fights its own battle against noise and degradation; some win, some lose. But the MMSE estimate wins the war 😉.
#iykuk #ImageRestoration

Diffusion Transformers with Representation Autoencoders by Boyang Zheng, et al (arxiv.org/abs/2510.116...)

Unexpected result: swapping the SD-VAE for a pretrained visual encoder improves FID, challenging the idea that encoders' information compression is not suited for generative modeling!

👏👏👏

We had an awesome #OMIBS2025

Thanks to all the lecturers, staff members, vendor faculty, sponsors, and participants for making this an amazing course year!

Introducing Latent-X — our all-atom frontier AI model for protein binder design.

State-of-the-art lab performance, widely accessible via the Latent Labs Platform.

Free tier: platform.latentlabs.com
Blog: latentlabs.com/latent-x/
Technical report: tinyurl.com/latent-X

👏