BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260422T143032Z
LOCATION:B314
DTSTART;TZID=America/New_York:20241117T090000
DTEND;TZID=America/New_York:20241117T173000
UID:submissions.supercomputing.org_SC24_sess741@linklings.com
SUMMARY:Machine Learning with Graphs in High-Performance Computing Environ
 ments
DESCRIPTION:The intent of this workshop is to bring together researchers, 
 practitioners, and scientific communities to discuss methods that utilize 
 extreme scale systems for learning graph data. This workshop will focus on
  the greatest challenges in utilizing high-performance computing (HPC) for
  machine learning with graphs and methods for exploiting extreme scale par
 allelism for data, computation, and model optimization. We invite research
 ers and practitioners to participate in this workshop to discuss the chall
 enges in using HPC for machine learning with graphs and to share the wide 
 range of applications that would benefit from HPC-powered machine learning
  with graphs.\n\nInvited Talk: Rethinking graph analytics benchmarks with 
 the AGILE workflow\n\nMarco Minutoli (Pacific Northwest National Laborator
 y (PNNL))\n---------------------\nAM-DGCNN: Leveraging Graph Attention Net
 works and Edge Attributes for Link Classification in Knowledge Graphs\n\nG
 raph-based representations are increasingly popular for storing and managi
 ng information through knowledge graphs, which capture entities and their 
 relationships. However, these knowledge graphs often suffer from incomplet
 e link information. To address this issue, link classification methods can
  be...\n\n\nDhroov Pandey and Tong Shu (University of North Texas)\n------
 ---------------\nAfternoon Break\n---------------------\nIRIS-GNN: Leverag
 ing Graph Neural Networks for Scheduling on Truly Heterogeneous Runtime Sy
 stems\n\nThe diversity of accelerators in computer systems poses significa
 nt challenges for software developers, such as managing vendor-specific co
 mpiler toolchains, code fragmentation requiring different kernel implement
 ations, and performance portability issues. To address these, the Intellig
 ent Runtime S...\n\n\nBeau Johnston (Oak Ridge National Laboratory (ORNL))
 ; Thibault de Boissiere (Canva Incorporated); and Mohammad Alaul Haque Mon
 il, Narasinga Rao Miniskar, Aaron Young, Seyong Lee, and Jeffrey S. Vetter
  (Oak Ridge National Laboratory (ORNL))\n---------------------\nLunch Brea
 k\n---------------------\nMDLoader: A Hybrid Model-Driven Data Loader for 
 Distributed Graph Neural Network Training\n\nScalable data management is e
 ssential for processing large scientific dataset on HPC platforms for dist
 ributed deep learning. In-memory distributed storage is preferred for its 
 speed, enabling rapid, random, and frequent data access required by stocha
 stic optimizers. Processes use one-sided or colle...\n\n\nJonghyun Bae (Go
 ogle); Jong Youl Choi, Massimiliano Lupo Pasini, Kshitij Mehta, and Pei Zh
 ang (Oak Ridge National Laboratory (ORNL)); and Khaled Z. Ibrahim (Lawrenc
 e Berkeley National Laboratory (LBNL))\n---------------------\nInvited Tal
 k: Advancing Graph AI: Tackling Efficiency, Application, and Explainabilit
 y\n\nYuede Ji (University of Texas at Arlington)\n---------------------\nA
 cceleration of Graph Neural Networks with Heterogenous Accelerators Archit
 ecture\n\nGraph Neural Networks (GNNs) have been used to solve complex pro
 blems of drug discovery, social media analysis, etc. Meanwhile, GPUs are b
 ecoming dominating accelerators to improve deep neural network performance
 . However, due to the characteristics of graph data, it is challenging to 
 accelerate GNN...\n\n\nKaiwen Cao (University of Illinois Urbana-Champaign
 , Hewlett Packard Labs); Archit Gajjar (Hewlett Packard Labs); Liad Gerstm
 an (Technion - Israel Institute of Technology); Kun Wu (University of Illi
 nois Urbana-Champaign); Sai Rahul Chalamalasetti (d-Matrix); Aditya Dhakal
 , Giacomo Pedretti, and Pavana Prakash (Hewlett Packard Labs); Wen-mei Hwu
  (University of Illinois Urbana-Champaign, NVIDIA Corporation); Deming Che
 n (University of Illinois Urbana-Champaign); and Dejan Milojicic (Hewlett 
 Packard Labs)\n---------------------\nClosing remarks\n\nJose Moseira (IBM
 )\n---------------------\nScalable and Consistent Graph Neural Networks fo
 r Distributed Mesh-based Data-driven Modeling\n\nThis work develops a dist
 ributed graph neural network (GNN) methodology for mesh-based modeling app
 lications using a consistent message passing layer. As the name implies, t
 he focus is on enabling scalable operations that satisfy physical consiste
 ncy via halo nodes at sub-graph boundaries. Here, con...\n\n\nShivam Barwe
 y, Riccardo Balin, Bethany Lusch, Saumil Patel, Ramesh Balakrishnan, and P
 inaki Pal (Argonne National Laboratory (ANL)); Romit Maulik (Pennsylvania 
 State University, Argonne National Laboratory (ANL)); and Venkatram Vishwa
 nath (Argonne National Laboratory (ANL))\n---------------------\nMorning B
 reak\n---------------------\nInvited Talk: Graphs in the LLM Era: Enabling
  Effective and Efficient LLM Ecosystems\n\nMaciej Besta (ETH Zürich)\n\nTa
 g: Artificial Intelligence/Machine Learning, Graph Algorithms, Scalable Da
 ta Mining\n\nRegistration Category: Workshop Reg Pass\n\nSession Chairs: S
 eung-Hwan Lim (Oak Ridge National Laboratory (ORNL)); José Moreira (IBM); 
 Catherine Schuman (University of Tennessee, Knoxville); and Richard Vuduc 
 (Georgia Institute of Technology)
END:VEVENT
END:VCALENDAR