2020 NSF Workshop on Smart Cyberinfrastructure


February 25-27, 2020, Hyatt Regency Crystal City, VA.


Rooms:

Plenary sessions in Conference Theater
Breakout 1 in Potomac I
Breakout 2 in Potomac VI
Breakout 3 in Potomac V

Day 1 – Tuesday, February 25, 2020

07:30 – 08:00 Registration/Breakfast
08:00 – 08:10

Valerio Pascucci, University of Utah & Steering Committee
Introductions, structure of the workshop, expected outcomes

08:10 – 08:40 William Miller, Science Advisor, NSF Office of Advanced Cyberinfrastructure (OAC)
Manish Parashar, Director, NSF Office of Advanced Cyberinfrastructure (OAC)
NSF Welcome and workshop introduction
08:40 – 10:00 Setting up the stage: Steering Committee (10 min presentations + 5 min questions)
Ilkay Altintas, San Diego Supercomputer Center
Improving AI/ML algorithms and associated data repositories on NSF Enabled Cyberinfrastructure
Ian Foster, The University of Chicago
Improve Efficient Use of the NSF Enable Cyberinfrastructure through AI/ML and smart technologies
Xinyu Zhao, University of Connecticut
Cyberinfrastructure requirements for AI/ML-based scientific applications
Michela Taufer, The University of Tennessee, Knoxville
Large collaborative initiatives and engagement of the international community
• Open discussion (20 min)
10:00 – 10:15 Coffee Break
10:15 – 10:45 Keynote talk 1, Amir Gholami, UC Berkeley
An Integrated Approach for Efficient Neural Network Design, Training, and Inference
10:45 – 11:45 Panel Discussion 1: AI/ML opportunities for a modernization of the NSF cyberinfrastructure from the edge to large facilities.
Organizer: Dan Stanzione
Panelists: Ian Foster, Salim Hariri, Jelena Mirkovic, Helen Gu
11:45 – 12:00 Organize breakout sessions
12:00 – 13:00 Lunch Break
13:00 – 13:15 Plenary talk 1, Karan Bhatia, Google
The Google Cloud
13:15 – 13:30 Plenary talk 2, Yoshio Tanaka, AI Bridging Cloud Infrastructure (ABCI)
Challenges for Smart Cyberinfrastructure in Making Society 5.0 a Reality
13:30 – 13:45 Keynote talk 2, Fleming Crim, Chief Operating Officer, National Science Foundation,
The NSF vision
13:45 – 14:00 Plenary talk 3, Bill Tang, Princeton Plasma Physics Laboratory
Features of an NSF Smart Cyberinfrastructure Roadmap Supporting Science Applications
14:00 – 14:15 Plenary talk 4, Mark Asch, University of Picardy,
Smart CI and the Real World
14:15 – 15:15 Breakout sessions start parallel discussions:
Breakout 1 in Potomac I: Cyberinfrastructure requirements for AI/ML-based scientific applications (lead Xinyu Zhao, co-leads/scribes Jose Fortes, Helen Gu)
Breakout 2 in Potomac VI: Development of the next generation physical (hardware) cyberinfrastructure tightly integrated with AI, ML and smart technologies (lead Dan Stanzione, co-leads/scribes Ian Foster, Steve Petruzza).
Breakout 3 in Potomac V: Improving AI/ML algorithms on NSF Enabled Cyberinfrastructure: (Lead: Ilkay Altintas, co-leads/scribes Michela Taufer, Timo Bremer)
15:15 – 15:30 Coffee break
15:30 – 16:00 Breakout sessions reconvene and complete answers to main charge questions
16:00 – 16:15 Breakout leads summarize findings and recommendations in a set of slides and a shared document.
16:15 – 17:00 Leads report to the entire workshop (15 min each group including questions)

Day 2 – Wednesday, February 26, 2020

07:00 – 08:00 Breakfast
08:00 – 08:10  Valerio Pascucci, University of Utah & Steering Committee
Summary of Day 1 / Goals of the Day 2
08:10 – 08:30  Plenary Talk 5, Nicola Ferrier, Argonne National Laboratory
AI@Edge: a Software-Defined Sensor Network
08:30 – 09:40  Lightning talks session 1
10 minutes each (5 slides + panel) 
09:40 – 10:00  Plenary Talk 6, Hank Hoffmann, The University of Chicago,
AI/ML methods for computer systems optimization
10:00 – 10:15  Coffee Break 
10:15 – 11:15  Lightning talks session 2
10 minutes each (5 slides + panel) 
11:15 – 12:15  Panel discussion 3: Improving AI/ML algorithms and associated data repositories on NSF Enabled Cyberinfrastructure:
Organizer: Ilkay Altintas,
Speakers: Peer-Timo Bremer, Yoshio Tanaka, Andrew Chien, Karan Bhatia
12:15 – 13:10  Lunch Break 
13:10 – 14:10  Panel Discussion 4: Effective Workforce Training, development of a community of
researchers and practitioners.
Organizer: Michela Taufer
Panelists: Abani Patra, Jose Fortes, Mark Asch, Morris Riedel, Carlos Costa
14:10 – 14:15  Organize breakout sessions 
14:15 – 15:15  Breakout sessions start parallel discussions:
Breakout 4 in Potomac V: AI, ML and smart technologies for data acquisition, experiments, and the edge of the cyberinfrsrtucture (lead Hank Hoffmann, co-lead/scribe Salim Hariri, Steve Petruzza)
Breakout 5 in Potomac I: Improving AI/ML associated data repositories on NSF Enabled Cyberinfrastructure (Lead Ilkay Altintas, co-lead/scribe Helen Gu, Timo Bremer)
Breakout 6 in Potomac VI, Effective Workforce Training, development of an international community of researchers and practitioners (Lead Michela Taufer, co-lead/scribe Terry Moore, Jose Fortes)
15:15 – 15:30  Coffee break 
15:30 – 16:00  Breakout sessions reconvene and complete answers to main charge questions 
16:00 – 16:15  Breakout leads summarize findings and recommendations in a set of slides and
a shared document. 
16:15 – 17:00  Leads report to the entire workshop (15 min each group including questions) 

Day 3 – Thursday, February 27, 2020  

07:00 – 08:00  Breakfast 
08:00 – 08:15  Valerio Pascucci, University of Utah & Steering Committee
Summary of Days 1-2 / Goals of the Day 3
08:15 – 08:35  Plenary Talk 7, Dan Stanzione, The University of Texas at Austin
Providing AI/DL/ML support in a Large Shared Facility: Perspective from TACC
08:35 – 08:55  Plenary Talk 8, Eliu Huerta, National Center for Supercomputing Applications
Convergence of AI and HPC for Big-Data Experiments 
08:55 – 09:15  Plenary Talk 9, Morris Riedel, Juelich Supercomputing Centre
AI in international initiatives
09:15 – 10:00 Final Breakout sessions: Feedback from community, assign responsibilities and start drafting report. 
10:00 – 10:15 Coffee break 
10:15 – 10:45  Continue Final Breakout sessions: Commit to writing assignments after workshop.
10:45 – 11:15 Keynote talk 3, Henry Kautz, NSF, Director, Division of Information and Intelligent Systems (CISE/IIS), National Science Foundation
An NSF Vision for AI Computational Infrastructure: From Tensor Clouds to the Edge
11:15 – 11:30  Presentation for draft summary findings and recommendations (steering committee)
11:30 – 12:00 Open discussion and feedback
12:00 – 12:15 Closing remarks, assignments for writing the report