I am delighted to announce that the camera-ready version of my new book, "Machine Learning: Advanced Topics", is finally available online for free at probml.github.io/book2 (The MIT Press @mitpress.bsky.social will publish the hard copy in 2023.)

Just out today, our work in mmWave+Depth metal detecting of concealed objects using sensor fusion, in MDPI Sensors! mdpi.com/1424-8220/24/1… Daniele Faccio Chaitanya Kaul QuantIC MDPI UofGARC UofG SciEng

Our work has made the cover of Sensors: AI Sensor Fusion of mmWave and Depth for Concealed Metal Detection- selected from 300 articles! mdpi.com/1424-8220/24/18 Daniele Faccio Chaitanya Kaul QuantIC UofGARC UofG SciEng

3. with pixelsmith, thanos and Marco Aversa develop a framework to achieve gigagpixel image generation on single gpu from existing diffusion models arxiv.org/abs/2406.07251

Pixelsmith is a new method to generate any-resolution (> 1024x1024) images using diffusion models! They upsample images to the desired resolution, add noise, and slowly denoise it with a patch-based guidance. It works on a single GPU! thanos-db.github.io/Pixelsmith/

GUESS WHO'S BACK BACK AGAIN ARCHIVE'S BACK TELL A FRIEND archive.org

"Proud to share our work with you all! 🎉 Check out our tool, give it a spin, and enjoy! Looking forward to connecting in person at NeurIPS – see you there! 👋 #NeurIPS2024 #researchpaper #GenerativeAI #generativeart

A quite literal game-changer- so exciting to watch this work develop into such a powerful tool! Congratulations Thanos Marco Aversa Chaitanya Kaul Daniele Faccio

Exciting news! 🚀 The code for Pixelsmith is now available! You can try it out here: github.com/Thanos-DB/Pixe…

Our latest work, Pixelsmith, accepted at NeurIPS 2024. Allows zero-shot image generation at gigapixel scale with Diffusion Models on a single RTX 3090 GPU.

Congratulations to Semiconductor, Daniele Faccio and everyone involved! 🙌 Some of the filming took place in the ARC and we really hope we can show this film in our atrium in the future. 📽️

Excited to finally share our latest work on diffusion as a confinement mechanism, guiding light around bends and with up to 100x higher transmission through completely opaque materials. Works also for heat and neutrons! Daniele Faccio UofGARC UofG SciEng nature.com/articles/s4156…

The work, published this morning in Nature Physics and reported in New Scientist institutions.newscientist.com/article/245418… sheds new light on waveguiding through opaque diffusive materials like resin (pictured) snow, fog and human tissue. Daniele Faccio Jack Radford Extreme Light UofG PhysAstro

The General Chairs @ger_ac83 Edmond Ho are delivering welcome speech!

'Is One GPU Enough? Pushing Image Generation at Higher-Resolutions with Foundation Models' was presented at NeurIPS Conference Led by Thanos & Marco Aversa, with Chaitanya Kaul Daniele Faccio and myself. Demos, paper and code at: thanos-db.github.io/Pixelsmith/ QuantIC Dotphoton

Please consider applying for this Ph.D. studentship with us!

We review Active Inference and how it could be applied to model the human-computer interaction loop. Active Inference and Human-Computer Interaction arxiv.org/abs/2412.14741 with John Williamson Sebastian Stein in the DIFAI ERC project DIFAI ERC Project difai-project.org

This is a great chance for a student in a computational interaction group or an active inference group to come and spend 3 months with us in Glasgow.