Publications

The Gemini 2.5 model family, advancing reasoning, multimodality, long context and agentic capabilities.

AlphaFold 3 — accurate prediction of the structure of complexes including proteins, nucleic acids, ligands and more. (Nature)

A compute-efficient multimodal model with recall and reasoning over millions of tokens of context.

Google DeepMind’s family of natively multimodal models across text, images, audio, video and code.

A framework (CDCD) for modelling categorical data such as text with continuous diffusion models.

Applying AlphaFold at scale to predict structures across the entire human proteome. (Nature)

AlphaFold — the deep learning system that solved the protein structure prediction problem. (Nature)

Provably convergent convex optimization using Deep Learning.

We use deep learning to perform bayesian inversion.

Task adapted reconstruction for inverse problems.

Learned Primal-Dual for DBT

We show how learning can be applied to proximal Primal-Dual schemes while still guaranteeing convergence.

Ultramicroscopy

Generalizing Wasserstein GAN with gradient penalty to Banach Spaces.

Compute the image-space Wasserstein loss using Sinkhorn iterations and use it for learning to solve inverse problems.

Using gradient boosting we scale learned iterative reconstruction to a very large scale inverse problem.

By learning in both reconstruction and data domain, we can improve image reconstruction.

Fully3D

A deep learning scheme for solving inverse problems that incorporates knowledge about the data formation process, improving both speed and quality compared to classical methods.

Fully3D

IFMBE Proceedings