Fifth National Research Platform (5NRP) Workshop
March 19 - 22, 2024
Co-located with the FABRIC Community Workshop - KNIT8
Full agenda HERE

Live streaming available:
Wednesday, 3/20 - https://youtube.com/live/JFrrP5gyjJU?feature=share

Thursday, 3/21 - https://youtube.com/live/5gVlb3QNdMc?feature=share
Friday, 3/22 - https://youtube.com/live/HrS00QFWmx4?feature=share
 

Writing a successful CC* Grant
Tutorial Lead: Jerry Sheehan, Salk Institute and Jason Zurawski, ESnet

This tutorial will explore best practices for developing NSF Campus Cyberinfrastructure proposals. Topics addressed will include the 2024 CC* Solicitation, Identifying science drivers, Best Practices for developing your proposal technology plan, Strategies to meet your 20% Resource Sharing requirements, and Tips to tell your Broader Impact story. At the conclusion of the tutorial, NSF Program Director from the Office of Advanced Cyberinfrastructure Kevin Thompson, will participate in a question-and-answer session with participants.

NRP, from AI to Networking Research to Education
Location: QI Atkinson Auditorium 

Frank Wuerthwein, UC San Diego

We will present the vision and status of NRP with particular focus on how NRP addresses three challenges.

 - The gap between those who have and those who can’t afford is getting wider.

-  Education increasingly requires significant cyberinfrastructure

- The end of Moore’s Law is leading to a proliferation of “architectures”, increasing the complexity of computing to the point where domain science adoption of  innovations is at risk

We will look at these challenges with AI as a lense. Thought these challenges are far more general.

Bringing AI to wireless sensing, communication, networking with NRP
Dinesh Bharadia, UC San Diego

Wireless radios can sense and communicate robustly in non-line of sight and inclement weather, enabling the entire era of mobile and AR/VR computing and interactions. This talk will present extensive data sets, deep learning models, and digital twins developed to enable wireless radios to provide sensing for autonomous cars, robotics, AR/VR sensing and content delivery, and 6G. In the era of generative AI, limited to only meaningfully processing large-scale text, image, and audio data due to abundant datasets, it can't penetrate these new forms of sensing; the lab's broader goal is to enable that in the next decade.

Learning Humanoid Robots
Xiaolong Wang, UC San Diego

Having a humanoid robot operating like a human has been a long-standing goal in robotics. The applications go much beyond moving boxes in the factory. The humanoid robot provides a generalized purpose platform to conduct diverse tasks we do in our daily lives. In this talk, we study learning-based approaches for both the mobility and manipulation skills of the humanoid robot, with the goal of generalization to diverse tasks, objects and scenes. With Reinforcement Learning and Sim2real transfer, we will first discuss how to perform whole-body control in humanoids to generate rich, diverse, and expressive motions. We then discuss what will be the foundation model for the robot to perceive the visual world and perform mobile manipulation in the wild. Finally, we will dive into the humanoid’s manipulator, the hands, and study tactile sensing for flexible manipulations.

National Data Platform (NDP) as a Research Resource for AI
IIkay Altintas, UC San Diego

Open and equitable access to scientific data is essential to addressing important scientific and societal grand challenges, and to research enterprise more broadly. This session discusses the importance and urgency of open and equitable data access, and introduces the vision and architecture of the National Data Platform, a recently launched project aimed at catalyzing an open, equitable and extensible data and service ecosystem for AI workflows.
 

NDP LLM-as-a-Service on NRP Tutorial
IIkay Altintas, UC San Diego

This tutorial offers an introduction to utilizing open Large Language Models (LLMs) as a service through the National Data Platform (NDP), focusing on their integration into various domain-specific application workflows. Participants will learn how to access and deploy LLM services through practical notebooks on NDP with hands-on examples running on NRP, navigating through setup, customization, and interaction steps. The technical depth of this tutorial session is balanced toward practical applications and accessibility to a broad audience to take advantage of models available on HuggingFace.

FABRIC Keynote
Location: QI Atkinson Auditorium
kc claffy, UC San Diego

The goal of the GMI project (Designing a Global Measurement Infrastructure to Improve Internet Security, or GMI3S) is to design a new generation of measurement infrastructure for the Internet, which will support collection, curation, archiving, and expanded sharing of data needed to advance critical scientific research on the security, stability, and resilience of Internet infrastructure.

Lunch/National Research Platform International Extensions, Including the Global Research Platform
Location: VROOM, Atkinson Hall (next to the Auditorium)

Bring lunches from Lobby

Agenda
A) Introduction to Session (Joe Mambretti) (12:15 – 12:25, 10 min)
B) NSF International Research Connections Program (Kevin Thompson) (12:25-12:35, 10 min)
C) AutoGOLE/SENSE (Tom Lehman) (12:35-12:45, 10 min)
D) Asia Pacific Research Platform (Jeonghoon Moon) (12:45-12:55, 10 min)
E) Ampath/AmLight (Julio Ibarra) (12:55-1:05, 10 min)
F) CENI International Testbed (Gauravdeep Shami) (1:05-1:15) (10 min) 

G) GNA-DiS Overview (Harvey Newman) (1:15-1:25, 10 min)
 

AI/ML Computing for the Laser Interferometric Gravitational Wave Observatory (LIGO)
Phil Harris, MIT

Deep learning algorithms have excelled in various domains. Despite this success, few deep-learning models have seen full end-to-end deployment in gravitational-wave searches, both in real-time and on archival data. In particular, there is a lack of standardized software tools for quick implementation and development of novel AI ideas. We address this gap by developing the ML4GW and HERMES libraries. We show how these libraries enhance efficiency and AI model robustness in the context of a broad range of gravitational wave analyses with an emphasis on real-time application, scalability to heterogeneous computing resources, and streamlining the training to deployment cycles for machine learning models. We present this work within the larger effort toward heterogeneous integration of large scale physics experiments and discuss where this work is going. Out of this scope, we present a possible model of next-generation computing for scientific experiments. 
 

Larry Smarr, UC San Diego

NRP storage and computing resources are being used to investigate how  AI/ML algorithms can be applied to a wide range of disciplines. I will give brief examples drawn from the research projects of a number of large NRP users from chemistry, physics, healthcare, drug discovery, robotics and manufacturing.

Expanding the AI/ML Workforce Through Coursework
Michael Farley, SDSU, Valerie Polichar, UC San Diego

Michael Farley, San Diego State's Chief Technology Research Officer, will share how SDSU has leveraged the NRPs "bring-your-own-resource" model to support instruction with their VERNE cluster. VERNE provides faculty an easy to use resource to teach a variety of courses that require access to advanced cyberinfrastructure.

Dr. Polichar will share how UC San Diego created a direct line from research to instruction by embracing vision and technology developed by elite researchers to build UCSD's Data Science/Machine Learning Platform (DS/MLP) for graduate and undergraduate instruction. Now a showcase for modern technology education, the DS/MLP is used by thousands of students every quarter for classes in every division, from Electrical Engineering to Visual Art to Public Health, instilling market-competitive skills in students across a broad range of disciplines.

Great Plains Network
Alex Hurt, UMissouri and Derek Weitzel, UNL - GP-Engine

This talk will present the latest research and teaching activities being performed on the National Research Platform in the Great Plains Network, which encompasses 8 universities in 6 great-plains states. This presentation will also include the activities and outreach performed as part of the GP-ENGINE project, which has expanded NRP usage throughout the GPN and added several advanced, high-performance GPU nodes as well as facilitated the addition of the first Grace Hopper Super-Chip to the National Research Platform.