ML4LMS

Socials

Tue, 26 Jul 2022 06:00:00 +0000

Socials Sponsored By Relation Therapeutics, VantAI, and Microsoft

Please note this event is exclusive to workshop attendees. Sign up via 🔗 THIS LINK st

Closing Remarks

Tue, 26 Jul 2022 06:00:00 +0000

Talk Title

Talk abstract

Rianne van den Berg

Tue, 26 Jul 2022 06:00:00 +0000

Generative Modeling for the Natural Sciences

In this talk will cover applications of generative models in different areas of the natural sciences. More details to come closer to the event.

Panel - Material Science

Tue, 26 Jul 2022 06:00:00 +0000

Material Sciences Panel

This panel brings together leading experts to explore the transformative intersection of Material Sciences and Machine Learning.

Our esteemed panellists include

David Pfau from DeepMind, renowned for his pioneering work fusion processes and materials;
Kevin Jablonka of the Helmholtz Institute for Polymers in Energy Applications. * Philippe Schwaller from EPFL, an authority on chemical reaction prediction using deep learning;
Sebastian Pattinson of Matta, University of Cambridge, whose research integrates advanced materials with AI for innovative solutions;
Rianne van den Berg, Principal Research Manager at Microsoft, who spearheads AI-driven advancements in material discovery.

Together, they will discuss recent breakthroughs, ongoing challenges, and future directions in utilising machine learning to accelerate material innovation, optimise properties, and discover new functionalities. This panel promises to provide valuable insights into how cutting-edge machine learning techniques are shaping the future of material sciences, fostering interdisciplinary collaboration, and driving technological advancements.

Coffee & Poster Session

Tue, 26 Jul 2022 06:00:00 +0000

Coffee Break Sponsored by Pfizer

Pfizer Inc. is an American multinational pharmaceutical and biotechnology corporation headquartered at The Spiral in Manhattan, New York City. The company was established in 1849 in New York by two German entrepreneurs, Charles Pfizer (1824–1906) and his cousin Charles F. Erhart (1821–1891).

Pfizer develops and produces medicines and vaccines for immunology, oncology, cardiology, endocrinology, and neurology. The company’s largest products by sales are the Pfizer–BioNTech COVID-19 vaccine ($11 billion in 2023 revenues), apixaban ($6 billion in 2023 revenues), a pneumococcal conjugate vaccine ($6 billion in 2023 revenues), palbociclib ($4 billion in 2023 revenues), and tafamidis ($3 billion in 2023 revenues). In 2023, 46% of the company’s revenues came from the United States, 6% came from Japan, and 48% came from other countries.

The company ranks 38th on the Fortune 500 and 39th on the Forbes Global 2000.

Guenter Klambauer

Tue, 26 Jul 2022 06:00:00 +0000

Towards Broad AI for Molecules and Drug Discovery

“Over the last decade, machine learning and Deep Learning methods have paved their way into all kinds of computational task for molecules. The molecular machine learning research community believes that it has made strong progress in * a) activity and property prediction, * b) representation learning and molecular modeling, * c) chemical synthesis and reaction prediction, and * d) generative models for molecules. But have we really made progress? QSAR models have been around since the 1960s and we might have only slightly increased predictive performance. Have these methods deserved the name ”Artificial Intelligence”? In this talk, we provide a perspective recent progress in molecular machine learning, on the essential properties that our AIs should have to make a difference, and steps towards such broad AIs.”

Burkhard Rost

Tue, 26 Jul 2022 06:00:00 +0000

Artificial Intelligence Deciphers the Code of Life Written in Proteins?

The objective of our group is to predict aspects of protein function and structure from sequence. The wealth of evolutionary information available through comparing the whole bio-diversity of species makes such an ambitious goal achievable. Our particular niche is the combination of evolutionary information (EI) with machine learning (ML) and artificial intelligence (AI). 30 years ago, the marriage of machine learning and evolutionary information (in the form of Multiple Sequence Alignments) allowed a breakthrough in secondary structure prediction. The same principle has been underlying all state-of-the-art predictions of protein structure and function and is also the root for the program that broke through in protein structure prediction, namely AlphaFold2.

Over the last two years, it has become possible to deep learn the language of life written in proteins through protein Language Models (pLMs). The information extracted is transfer learned to supervise learn protein prediction with annotations. I will present three particular new methods predicting protein structure (1D: secondary structure, membrane regions, & disorder, 2D: inter-residue distances/contacts, 3D: co-ordinates) and protein function (sub-cellular location, binding residues, GO terms), and the effects of sequence variation using pLMs. These embeddings allow for some applications to reach for others to surpass the state-of-the-art without using evolutionary information.

Crucial in all of this is the understanding of the AI and the control of database bias. For both computational biology could serve as a sandbox to prepare more sensitive applications of AI in society.

Coffee Break

Tue, 26 Jul 2022 06:00:00 +0000

Coffee Break Sponsored by SandBoxAQ

In the tech and data intelligence worlds, a sandbox is where innovation is born. It’s a place where the brightest free-thinking minds from across disciplines come together to reimagine what’s possible. A collaborative environment where the whole is infinitely greater than the sum of the parts.

At SandboxAQ, this forward-looking vision is core to everything we do. It’s how we became who we are and it’s how we know our solutions can shift the way your business competes in tomorrow’s marketplace. As the world enters the third quantum revolution, AI + Quantum software will address significant business and scientific challenges. ‍ SandboxAQ is an enterprise SaaS company, providing solutions at the nexus of AI and Quantum technology (AQ) to address some of the world’s most challenging problems. The company’s core team and inspiration formed at Alphabet Inc., emerging as an independent, growth-capital-backed company in 2022.

Contributed Talks

Tue, 26 Jul 2022 06:00:00 +0000

Contributed Talk III

Generative acceleration of molecular dynamics simulations for solid-state electrolytes

Presented by: Juno Nam

We introduce LiFlow, a generative acceleration framework designed for efficiently simulating diffusive dynamics in solids, particularly lithium-based solid-state electrolytes (SSEs). LiFlow consists of two components: Propagator and Corrector, which utilize a conditional flow matching scheme to predict atomic displacements and perform denoising, respectively. Our model achieves a Spearman’s rank correlation of approximately 0.7 for the lithium mean squared displacement (MSD) on test set based on composition and temperature splits and offers a substantial speedup compared to reference molecular dynamics (MD) simulations using machine learning interatomic potentials (MLIPs). This framework facilitates high-throughput virtual screening for electrolyte materials and holds promise for the optimization of the kinetic properties of crystalline solids.

Contributed Talk IV

A Recipe for Charge Density Prediction

Presented by: Julia Balla

In density functional theory, charge density is the core attribute of atomic systems from which all chemical properties can be derived. Machine learning methods are promising in significantly accelerating charge density prediction, yet existing approaches either lack accuracy or scalability. We propose a recipe that can achieve both. In particular, we identify three key ingredients: (1) representing the charge density with atomic and virtual orbitals (spherical fields centered at atom/virtual coordinates); (2) using expressive and learnable orbital basis sets (basis function for the spherical fields); and (3) using high-capacity equivariant neural network architecture. Our method achieves state-of-the-art accuracy while being more than an order of magnitude faster than existing methods. Furthermore, our method enables flexible efficiency-accuracy trade-offs by adjusting the model/basis sizes.

Frank Noe

Tue, 26 Jul 2022 06:00:00 +0000

Protein Thermodynamics and Kinetics with Deep Learning

Talk abstract coming soon