National Artificial Intelligence Research Resource (NAIRR ) Software Workshop

Dec 3, 2024, 8:00 AM → Dec 4, 2024, 6:00 PM America/Chicago

TCS Conference Center (Argonne National Laboratory)

Dhabaleswar (DK) Panda (Ohio State University), Michael Papka (University of Illinois Chicago / Argonne National Laboratory), NAIRR Workshop

Workshop Charge: The National AI Research Resource (NAIRR) Task Force has called for democratizing access to artificial intelligence, emphasizing the urgent need to break down barriers to AI capabilities. With the backing of the National Science Foundation’s Office of Advanced Cyberinfrastructure and the Department of Energy’s Advanced Scientific Research Computing program, we are organizing a workshop to help define an AI software stack that supports the following NAIRR goals:

1. Spurring Innovation: Facilitating cutting-edge AI developments that foster scientific advancements.
2. Increasing Diversity of Talent: Broadening participation across various demographics enriches the AI community with diverse perspectives.
3. Improving Capacity: Enhancing the ability of individuals and organizations to utilize AI effectively and efficiently.
4. Advancing Trustworthy AI: Promoting AI systems that are ethical, transparent, and trustworthy.

Workshop Objectives: The workshop aims to define a comprehensive AI software stack that includes computer programs, training and inference frameworks, libraries, user interfaces, debuggers, and performance tools. This stack should be accessible to a broad audience and adaptable to the evolving needs of the scientific community over a 2-5-year timescale. Key areas of focus will include:

• Real-Time Data Analysis: Addressing the needs for real-time processing and decision-making using AI in scientific applications.
• Privacy and Security: Ensuring AI applications meet rigorous privacy and security standards.
• Foundation Models and Libraries: Incorporating cutting-edge models and libraries, including mixed precision and emerging AI hardware platforms.
• Software Portability: Enabling software to be portable across diverse hardware platforms, considering variability in hosting needs, data formats, standards, and privacy policies.
• Data and Model Curation: Integrating robust protocols for the collection, validation, annotation, and maintenance of datasets and models within the software stack to ensure high-quality, relevant, and up-to-date resources for AI research and development.

The workshop will:

• Identify AI Software Needs: Provide a detailed understanding of AI software requirements across various domains and user groups.
• Leverage Existing Software Stacks: Incorporate proven solutions from academia, laboratories, and industry to enhance the NAIRR stack’s robustness.
• Set Priorities for Development: Establish clear priorities for developing and deploying the software stack, focusing on critical components and functionalities.
• Define Integration Processes: Outline procedures for integrating new software into the NAIRR stack, ensuring its continued relevance and effectiveness.
• Promote Ethical AI Practices: Emphasize the importance of ethical, transparent, and trustworthy AI systems in scientific research.

 

 

NAIRR Pilot

NAIRR Report

Securing The Open-Source Software Ecosystem

Tuesday, December 3

8:00 AM Breakfast and Check-in

1 Workshop Overview and Objectives

9:00 - 9:05 Introductions (DK Panda and Michael Papka)
9:05 - 9:10 NAIRR Software Workshop Goals and Outcomes (Sheikh Ghafoor)
9:10 - 9:30 NAIRR Overview (Katie Antypas)

2 Domain AI Talks: AI in Practice‚ AI What is Missing

Five 10-minute talks from domain scientists on AI usage, gaps, and challenges.

09:30 - 09:40 Climate (Troy Arcomano)
09:42 - 09:52 Medical (Aldo Badano)
09:54 - 10:04 Instruments/Experiments (Mathew Cherukara)
10:06 -10:16 Biology (Gautham Dharuman)
10:18 - 10:28 Environment (Carl Boettiger)

3 Surveys

10:30 - 10:40 NERSC Survey Results (Wahid Bhimji)
10:40 - 10:50 NAIRR Software Survey Results (Murali Emani)

10:50 AM Break

Reconvene in breakout rooms (1416, 1404, 1405, 1406, 1407)

Breakout Session 1: Software Needs for NAIRR Pilot (All Rooms)

Rooms 1416, 1404, 1405, 1406, 1407

4 Current Software, Tools, and Gaps - Training and Inference

Room 1416 - Lead: Diana McSpadden Scribe: Khalid Hossain
Room 1404 - Lead: Zhao Zhang Scribe: Krishna Teja Chitty-Venkata
Room 1405 - Lead: Michael Papka Scribe: Shilpika
Room 1406 - Lead: DK Panda Scribe: Shivam Barwey
Room 1407 - Lead: Nicola Ferrier Scribe: Archit Vasan

5 Current Software, Tools and Gaps - Data Management and Storage

6 Current Models, Datasets and Gaps

7 Current Accessibility & Usability and Gaps

8 Current Software, Tools, and Gaps - Security & Privacy

9 Summary of Breakout 1

Each of the five breakout teams from Breakout 1 will present the highlights of their discussions. Each team is allotted 5 minutes for their report-out, which allows approximately one minute per topic.

1:00 PM Lunch (Networking)

Reconvene in room 1416

10 Cosmology Meets AI: Roadmapping the Final Frontier

Speaker: Salman Habib (Argonne National Laboratory)

Parallel Breakout Session 2: Key Software Features

Rooms 1416, 1404, 1405, 1406, 1407

11 Room 1404 - Privacy & Security

Lead: Anita Nikolich
Scribe: Khalid Hossain

12 Room 1405 - Real-Time Analysis

Lead: Wahid Bhimji
Scribe: Shilpika

13 Room 1406 - Emerging Technologies

Lead: Feiyi Wang
Scribe: Nwamaka Okafor

14 Room 1407 - Usability

Lead: Ilkay Altintas
Scribe: Shivam Barwey

15 Room 1416 - Portability and Performance

Lead: Murali Emani
Scribe: Krishna Teja Chitty-Venkata

4:00 PM Break

Reconvene in room 1416

16 Summary of Breakout 2

Each of the five breakout teams from Breakout 2 will present the highlights of their discussions. Each team is allotted 5 minutes for their report-out.

17 Day 1 Recap and Discussion

Summary of key insights and preview of Day 2

5:00 PM Return Shuttles to Crowne Plaza

Shuttles will depart at 5:00 p.m. and 5:30 p.m.

Wednesday, December 4

8:00 AM Breakfast

Participant breakfast and networking

18 Welcome Back and Recap of Day 1

Review of Day 1 key takeaways

Parallel Breakout Session 3: Software Adoption & Deployment Challenges

19 Room 1404 - Deployment Challenges

Lead: Ilkay Altintas
Scribe: Krishna Teja Chitty-Venkata

20 Room 1405 - Adoption Barriers

Lead: Wahid Bhimji
Scribe: Khalid Hossain

21 Room 1406 - Outreach

Lead: Ian Kash
Scribe: Nwamaka Okafor

22 Room 1407 - Software Carpentry

Lead: Lois Curfman McInnes
Scribe: Shilpika

23 Room 1416 - Design Challenges

Lead: Nathan Tallent
Scribe: Archit Vasan

10:45 AM Break

Reconvene in room 1416

24 Summary of Breakout 3

Each of the five breakout teams from Breakout 3 will present the highlights of their discussions. Each team is allotted 5 minutes for their report-out.

Parallel Breakout Session 4: Addressing Short-and Long-Term Objectives (All Rooms)

25 2-Year Objectives

Room 1416 - Lead: Murali Emani Scribe: Khalid Hossain
Room 1404 - Lead: Shantenu Jha Scribe: Krishna Teja Chitty-Venkata
Room 1405 - Lead: Diana McSpadden Scribe: Shilpika
Room 1406 - Lead: Anita Nikolich Scribe: Shivam Barwey
Room 1407 - Lead: Nicola Ferrier Scribe: Archit Vasan

26 3-5 Year Objectives

27 NSF Efforts

28 Diverse Ecosystem

29 Inclusive Communities

30 Summary of Breakout 4

Each of the five breakout teams from Breakout 4 will present the highlights of their discussions. Each team is allotted 5 minutes for their report-out, which allows approximately one minute per topic.

1:15 PM Lunch (Networking)

31 Action Plan for Post-Workshop Process

Develop roadmap for reporting and stakeholder engagement

3:30 PM Break

32 Final Thoughts and Next Steps

Summary of key takeaways and final remarks

33 Closing Remarks and Adjournment

Wrap-up and thank you

5:00 PM Return Shuttles to Crowne Plaza