Skip to the content.


Thanks to the organizers, presenters, participants and sponsors. The workshop had 102 registrations (73-ORNL and 29-non-ORNL), 33 lightning talks, and 39 poster submissions.

The US Department of Energy (DOE) launched the Advancements in Artificial Intelligence for Science Funding Opportunity Program to accelerate the development and application of AI in the DOE Office of Science Advanced Scientific Computing Research (ASCR) program. The focus of ASCR research and development investments is on the underlying approaches for AI-enhanced scientific and engineering capabilities and to significantly transform the scientific method for accelerated discovery and innovation. It is therefore critical to enable trustworthy AI for scientific applications advancing human understanding and addressing national needs.

⤾ Back to top

Aims and Scope

Organized by Oak Ridge National Laboratory (ORNL), the lab is hosting a first workshop across projects funded by the AI4Science program. The workshop will provide an opportunity for participants to share their research, discuss challenges, and foster collaborations across areas. The workshop will also provide an opportunity for participants to engage with the broader scientific community at ORNL.

The workshop will feature keynote talks, project presentations, lightning talks/panels, and a poster session.

⤾ Back to top

Call for Lightning Talks and Posters

We invite people to register in-person with the option to submit lightning talk (3-5 minutes) and/or a posters in the areas aligned with the AI4Science program. Note: The lightning talk slots are full, but you can still submit a poster.

Other areas are also welcome impacting AI for science.

⤾ Back to top

Registration

Registration Link including Talk/Poster submissions: Click here to Register

To attend the workshop at ORNL in person please register by the following deadlines:

We encourage early registration for in-person participation due to the venue’s capacity

⤾ Back to top

Agenda

The workshop will be held at the JICS Auditorium, Bldg 5100, ORNL Main Campus, Oak Ridge, TN. Registration opens at 7am, and the workshop is scheduled from 9am-6pm.

Time Session Chair
Morning    
7:00am-9:00am Visitor registration in badge office 5200, Hang posters in JICS 5100  
     
9:00am-9:05am Welcome and Program Info William F Godoy, ORNL
9:05am-9:10am Opening Remarks Michael L Parks, ORNL CSMD Director
9:10am-10:00am Keynote Jack Dongarra, ORNL/UTK, 2021 Turing Award Winner
10:00am-10:15am Morning Break  
10:15am-11:45am Lightning Talks/Panel Session 1 Chair: Arjun Guha, NU
Working Lunch    
11:45am-12:45pm AI4Science projects overview Chair: Jeffrey S Vetter, ORNL
Afternoon    
12:45pm-1:15pm Break/Frontier tour and photos Chair: Scott Atchley, ORNL
1:15pm-2:30pm Lightning Talks/Panel Session 2 Chair: Pedro Valero-Lara, ORNL
2:30pm-2:45pm Break  
2:45pm-3:55pm Lightning Talks/Panel Session 3 Chair: William F Godoy, ORNL
3:55pm-4:00pm Final remarks William F Godoy, ORNL
4:00pm-6:00pm Poster Session JICS lobby  

⤾ Back to top

Organizers

Listed as Name, Institution - AI4Science Project(s)

Chairs:

Committee:

⤾ Back to top

Sponsors

⤾ Back to top

Lightning Talks Session 1

Title Presenter Institution
Democratizing AI: Open-source Scalable LLM Training on GPU-based Supercomputers Abhinav Bhatele University of Maryland
Distributed Data Store for Scalable Training Jong Choi ORNL
Pre-training retrieval models for code Tom Goldstein University of Maryland
Towards Efficient Pre-training via Low-Rank Tensor Optimization Zheng Zhang University of California - Santa Barbara
E4S: A platform for HPC-AI Sameer Shende University of Oregon/ParaTools, Inc.
Scalable Training of Trustworthy and Energy-Efficient Predictive Graph Foundation Models for Atomistic Materials Modeling: A Case Study with HydraGNN Massimiliano Lupo Pasini ORNL
Energy-Efficient Training for Large-Scale Vision Transformer Foundation Models Xiao Wang ORNL
Low-level forces are all you need…with multi-fidelity learning Richard Messerly ORNL
ChatHPC Pedro Valero-Lara ORNL

⤾ Back to Agenda

Lightning Talks Session 2

Title Presenter Institution
Flexible brain-inspired hybrid analog-spiking neuronal network computation in energy efficient neuromorphic hardware Christoph Kirst University of California, San Francisco
Graph-Based Problem-Solving on Neuromorphic Systems Catherine Schuman The University of Tennessee
Optimizing Graph-Based Neuromorphic Systems with Bayesian Optimization Shay Snyder George Mason University
Graph-Based Neuromorphic Systems with Hyperdimensional Computing Shay Snyder George Mason University
Rethinking Temporal Cross-Validation in Multivariate Subsequent Anomaly Detection Pablo Moriano ORNL
Scientific Hypothesis Generation Standing on the Shoulders of Literature-Based Discovery Tirthankar Ghosal ORNL
Privacy-Preserving Federated Learning for Foundation Models Kibaek Kim Argonne National Laboratory
Randomized linear gradient compression with applications to private training and uncertainty quantification Thomas Flynn Brookhaven National Laboratory
Multilevel Projection Algorithms for Handling Heterogenous Differential Privacy Ole Kotevska ORNL
Federated scientific machine learning for approximating functions and solving differential equations with data heterogeneity Lu Lu Yale University

⤾ Back to Agenda

Lightning Talks Session 3

Title Presenter Institution
Data-Driven Population Health Surveillance at Scale for Pandemic Readiness Adam Spannaus ORNL
ESGF-LLM: Building a Domain-Specific LLM for Earth System Data Elias Massoud ORNL
AI meets the Data Lifecycle Olga Kuchar ORNL
An Ensemble Score Filter for Tracking High Dimensional Nonlinear Complex Dynamical Systems Feng Bao Florida State University
Extracting Properties of Charged Polymers from Scattering Using Machine Learning Lijie Ding ORNL
MATEY: a multiscale adaptive turbulence foundation model Pei Zhang ORNL
Mechanisms and Networks and AI (Oh My!): LLMs and Agents for Systems Biology Dan Jacobson ORNL
Towards Semantics Lifting for Scientific Computing: A Case Study on FFT Naifeng Zhang Carnegie Mellon University

⤾ Back to Agenda

AI4Science Projects Overview Session

Title Presenter Institution
Projects Overview Introduction Jeffrey Vetter (moderator) ORNL
DURBAN: Enhancing Performance Portability in HPC Software with Artificial Intelligence Keita Teranishi ORNL
Ellora Productive AI-Assisted HPC Software Ecosystem Harshitha Menon LLNL
DyGenAI: Dynamic Generative AI for Prediction and Control of High-Dimensional Nonlinear Complex Systems Guannan Zhang ORNL
Privacy-Preserving Federated Learning for Science: Building Sustainable and Trustworthy Foundation Models Kibaek Kim ANL
ENGAGE: (E)nergy-efficient (N)ovel Al(g)orithms and (A)rchitectures for (G)raph L(e)arningres Guojing Cong ORNL
FedNeMO: Physics-Informed and Energy-Aware Federated Learning of Neural Multi-Operator Learners as Scientific Foundation Models Lu Lu Yale University

⤾ Back to Agenda