José Miguel Hernández Lobato probabilistic machine learning methods for scientific applications.

 José Miguel Hernández Lobato is indeed a Professor at the University of Cambridge who works on machine learning and its applications, including to molecular dynamics. Here are some key points about his research related to deep generative AI for molecular dynamics:

1. Hernández Lobato leads the Machine Learning for Science and Technology group at Cambridge, which develops probabilistic machine learning methods for scientific applications.

2. His group has worked on using deep generative models like variational autoencoders (VAEs) to model molecular systems and accelerate molecular dynamics simulations.

3. Some of their notable work includes:

   - Developing "Boltzmann generators", which are deep generative models trained to sample from equilibrium distributions of molecular systems, allowing efficient exploration of configuration space.

   

   - Using normalizing flows and other generative models to learn coarse-grained representations of molecular systems.

   

   - Applying variational inference techniques to learn force fields for molecular dynamics from quantum mechanical data.

4. They have applied these methods to problems like protein folding, drug discovery, and materials design.

5. Hernández Lobato collaborates with chemistry and materials science researchers to apply machine learning to molecular modeling challenges.

6. His work aims to combine the strengths of physics-based molecular dynamics with data-driven machine learning approaches.

7. The group develops open-source software implementations of their methods to facilitate adoption by the scientific community.

While not exclusively focused on molecular dynamics, Hernández Lobato's research represents some of the leading work on applying deep generative models and probabilistic machine learning to molecular simulation and related scientific domains. His group is at the forefront of integrating AI techniques with traditional computational chemistry approaches.

Citations:

[1] https://ppl-ai-file-upload.s3.amazonaws.com/web/direct-files/11431419/7487cc95-5fe5-4c3a-a0f8-808ae6476563/GenerativeMap_InfoVis_2019.pdf