hpc-ch forum on AI/ML Services for Science

Thursday 31 Oct 2024, 09:30 → 17:00 Europe/Zurich

CSCS Conference Room

Maria Grazia Giuffreda (CSCS), Pablo Fernandez (CSCS)

Description

As the landscape of scientific research evolves, so do the demands on High-Performance Computing (HPC) centers. Today, integrating Machine Learning (ML) and Artificial Intelligence (AI) services into HPC offerings is becoming increasingly essential. However, this integration comes with its own set of technical challenges and questions that need to be addressed.

Join us for an insightful discussion as we delve into the intersection of High-Performance Computing and Machine Learning, exploring the challenges and opportunities in providing ML/AI services for scientific research. Whether you're a seasoned HPC professional, a data scientist, or an AI enthusiast, this forum promises to offer valuable insights and perspectives from experts in the field. We invite both speakers and attendees to contribute to this important conversation shaping the future of scientific computing.

Key Questions

1. Challenges in Offering AI/ML Services:
1. What are the unique challenges HPC centers face in providing AI/ML services alongside traditional HPC capabilities?
2. How do we ensure scalability and efficiency in AI/ML workflows within HPC infrastructures?
2. User Expectations:
1. What are the expectations of users regarding AI/ML services within HPC environments?
2. How can HPC centers tailor their offerings to meet these expectations effectively?
3. Concrete Machine Learning Workflows:
1. What are the typical workflows involved in integrating ML/AI with HPC resources?
2. How can these workflows be optimized for performance and reliability?
4. Tools for AI/ML Services:
1. What tools and frameworks are currently available for providing AI/ML services in HPC settings?
2. Are there any gaps in tooling that need to be addressed to better support AI/ML integration?
5. Data Requirements and Considerations:
1. What specific requirements does ML/AI integration impose on data storage, access, and management within HPC centers?
2. How can HPC centers navigate the complexities of handling diverse data types and sources?

 

10:00 → 10:30 Welcome Coffee & Registration

10:30 → 10:45 Welcome and Introduction

Speakers: Maria Grazia Giuffreda (CSCS), Michele De Lorenzi (CSCS), Pablo Fernandez (CSCS)

Welcome_1.jpg

Welcome_2.jpg

Welcome_3.jpg

10:45 → 11:30 AI for Climate Data Generation, Assimilation, and Modeling

This presentation explores the convergence of Artificial Intelligence (AI) and climate modeling for high-resolution (km-scale) climate prediction.

We will discuss the use of accelerated km-scale simulations to generate synthetic climate data, enabling the training of even more sophisticated climate models. We'll delve into the integration of observational data from diverse sources – weather stations, satellites, and airborne platforms – through diffusion models, a technique commonly used for image generation.

To address the exascale challenge posed by climate data volume, we will present an AI-based compression method utilizing deep neural networks for efficient data representation. This approach allows for near-lossless reconstruction of the original data for subsequent analysis. We posit that this synergistic approach, combining high-fidelity data generation with advanced data assimilation techniques, has the potential to significantly improve the accuracy of climate predictions. This, in turn, can inform policy decisions and guide societal responses to climate change.

Speaker: Torsten Hoefler (ETH Zurich)

hoefler-climate-ai-hpc-ch.pdf

Picture 1a.png

Picture 1.jpg

11:30 → 11:50 Accelerating Access to Data and Research Collaborations through Trusted Research Environments (TREs)

Speaker: Jean-Louis Raisaro (CHUV)

20241031 CSCS-Seminar_RAISARO.pdf

Pictures 2.png

11:50 → 12:20 An Overview of the "Alps" Research Infrastructure

Speaker: Maxime Martinasso (CSCS)

HPCCH_AlpsInfra_Oct2024.pdf

Picture 3.jpg

12:20 → 13:05 Lunch and Networking

13:05 → 13:35 Generative DNNs at Unibas: imagination, expectations, and infrastructure

We will discuss the landscape of use cases and potential solutions for generative deep neural networks applications in research, administrative processes, and teaching.

Speaker: Geoffrey Fucile (University of Basel)

HPC_CH_2024_Fucile.pdf

Picture 4.jpg

13:35 → 14:05 Benchmarking AlphaFold2 on the Euler Cluster

AlphaFold2 (AF2) is an ML-based software that revolutionized the structural biology field by folding protein sequences in 3D structures with accuracy that, in some cases, reached the experimental one. Consequently, AF2 has become a very popular software on the ETH Euler cluster. It has users with IT levels ranging from beginner to expert trying to request resources to fold their own protein sequences. We have conducted a wide range of tests to derive clear guidelines for requesting resources on Euler for AF2, avoiding the waste of CPU and GPU resources. In this presentation I will showcase how we managed to satisfactorily accommodate AF2 usage both for the users and us as the HPC provider.

Speaker: Nadia Marounina (ETH Zurich)

HPC-CH_31_Oct_2024.pdf

Picture 5a.png

Picture 5.png

14:05 → 15:00 Community Development

The session will be dedicated to:

• Introducing Our New Community Members: Agroscope and Idiap Research Institute
• Planning the hpc-ch forums for 2025

CSCS.jpg

Idiap-HPC-CH.pdf

IMG_2247.jpg

Picture 7.jpg

Picture 8a.jpg

Picture 9.jpg

15:00 → 15:10 Introduction to UNICC

The United Nations International Computing Centre (UNICC) has over 50 years of experience as the largest strategic partner for digital solutions and cybersecurity within the United Nations system. We design and deploy transformational digital tools and programmes to support over 90 partners in fulfilling their mandates. With our world-class Cybersecurity Centre of Excellence, data and analytics practice and an array of platform, software and infrastructure solutions, UNICC serves the entire of the UN family and other international organizations with similar missions and values for the benefit of the world. Join us in learning more about our continuous commitment to delivering innovative, forward-looking and reliable system-wide solutions in line with the UN Secretary-General’s Strategy on New Technologies, the UN Secretary-General’s Roadmap for Digital Cooperation and the Common Agenda.

Speaker: Gabriel Galati (United Nations International Computing Centre)

Picture 10.jpg

UNICC Presentation for HPC-CH Forum 31 Oct 2024.pdf

15:10 → 15:20 Chief Data & AI

The UNICC AI Hub is committed to enhancing Predictive and Generative AI capabilities to support UN partners. By fostering strategic partnerships and adopting a cloud-first approach, UNICC aims to ensure robust data management, ethical compliance, and optimal model performance. This initiative addresses key technical challenges using High-Performance Computing (HPC), promoting a transparent and responsible AI governance framework. Join us in exploring how these advancements can drive impactful solutions for the UN community.

Speaker: Anusha Dandapani (United Nations International Computing Centre)

15:20 → 16:20 Guided Tour of CSCS Machine Room

16:20 → 16:30 Farewell and End of the Meeting