BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260202T201337Z
LOCATION:Second Floor Atrium
DTSTART;TZID=America/Chicago:20251118T080000
DTEND;TZID=America/Chicago:20251118T170000
UID:submissions.supercomputing.org_SC25_sess537@linklings.com
SUMMARY:Poster Presentations (Interactive Research e-Posters and Doctoral 
 Showcase)
DESCRIPTION:Exploring Efficient Deep Learning Training on AI Accelerators\
 n\nThe computational and memory demands of DNN training have grown with th
 e size of AI models in recent years. To address these demands, popular acc
 elerators (i.e., GPUs) must find novel ways to reduce memory utilization s
 ince their memory capacity is on the scale of tens of GB. Other companies 
 have un...\n\n\nMilan Shah (North Carolina State University)\n------------
 ---------\nImproving Collective Aggregation for HPC and AI Workloads\n\nHi
 gh performance computing (HPC) applications generate massive volumes of da
 ta, placing sustained pressure on parallel file systems (PFS) that face li
 mited bandwidth and resource contention. While file-per-process I/O allows
  lock-free access, reducing stripe contention, it creates excessive metada
 ta...\n\n\nMikaila Gossman (Clemson University)\n---------------------\nDe
 signing GPU-Aware Collective Communication for Heterogeneous Clusters with
  Diverse GPUs and Interconnects\n\nGPU-accelerated HPC and deep learning w
 orkloads now operate at scales of tens to thousands of GPUs, making collec
 tive communication a dominant cost. Applications such as Amber, heFFTe, an
 d distributed LLM training require frequent synchronization and exchange o
 f large data partitions. At the same ti...\n\n\nChen-Chun Chen (The Ohio S
 tate University)\n---------------------\nAIDRIN: A Comprehensive Toolset f
 or Automating Data Preparation for AI\n\nHigh-quality, ethically-governed,
  and efficiently structured data is important for effective AI. However, o
 rganizations often lack a unified method to assess whether datasets are re
 ady for AI modeling. AIDRIN (AI Data Readiness Inspector) provides a compr
 ehensive, multi-pillar framework that quantif...\n\n\nKaveen Hiniduma (The
  Ohio State University), Jean Luca Bez (Lawrence Berkeley National Laborat
 ory (LBNL)), Ravi Madduri (Argonne National Laboratory (ANL)), and Suren B
 yna (The Ohio State University)\n---------------------\nAnalytics4X: Gener
 al-Purpose Framework for Analysis and Optimization of HPC Data Movement\n\
 nAs scientific applications tackle more complex problems, data movement ha
 s also grown in complexity to the point of slowing execution time and comp
 romising time-to-solution, hindering the pace of scientific discovery. In 
 this work, we claim that, to continue to accelerate scientific discovery i
 n the...\n\n\nIan Lumsden (University of Tennessee, Knoxville)\n----------
 -----------\nHeterogeneous HPC Compute Continuum: A Roadmap for Workflow M
 apping and Scheduling From Sensor to Supercomputer\n\nEfficient workload m
 apping and scheduling in heterogeneous HPC environments connecting from Io
 T, edge devices to cloud is essential for optimizing resource use, reducin
 g makespan, and ensuring adaptability. This research explores advanced sol
 utions addressing mapping and scheduling by investigating ...\n\n\nAasish 
 Sharma (Georg-August-Universität Göttingen, Gesellschaft für wissenschaftl
 iche Datenverarbeitung mbH Göttingen)\n---------------------\nCyberinfrast
 ructure-Driven Spatial Decision-Making Support: Addressing Participatory C
 ollaboration and Spatiotemporal Heterogeneity\n\nSpatial decision support 
 systems (SDSS) are pivotal in resolving complex geospatial challenges but 
 face critical limitations in harmonizing conflicting objectives, capturing
  behavioral heterogeneity, and enabling efficient large-scale data process
 ing. Besides, a central challenge is that current Geo...\n\n\nZhenlei Song
  (Texas A&M University)\n---------------------\nSketch-Based Algorithmic F
 rameworks for Genome-Scale Mapping\n\nSketching is a widely used class of 
 techniques aimed at generating compact representations of longer biologica
 l sequences. Instead of comparing sequences, sketches allow us to sample f
 rom a subspace of k-mers and use those samples for comparison, saving both
  time and memory in the end application. O...\n\n\nTazin Rahman (Washingto
 n State University)\n---------------------\nWhy Read It All? Just Read Wha
 t Matters: A Path to Faster Scientific Visualization\n\nAs HPC simulations
  generate ever-larger datasets, reducing the volume of data that must be l
 oaded into compute node memory for analysis has become essential for unloc
 king insights efficiently. In-storage analysis achieves this by processing
  data directly at the storage servers, allowing them to retu...\n\n\nQing 
 Zheng, Brian Atkinson, Jason Lee, and Gary Grider (Los Alamos National Lab
 oratory (LANL))\n---------------------\nViral Pneumonia Disease Classifica
 tion with Machine Learning Techniques\n\nPneumonia is a dreadful condition
  that is the primary cause of death globally for individuals of all ages, 
 but it is especially dangerous for small children who are younger than fiv
 e. The radiological results obtained from an X-ray could lead to mistakes,
  incorrect diagnoses, and unnecessary delays....\n\n\nRacheal Shade Akinbo
 , Olabode Olatubosun, and Emmanuel Ibam (Federal University of Technology 
 Akure Nigeria)\n---------------------\nOAAgent: A Multimodal LLM Agent Cli
 nical Assistant for Precision Osteoarthritis Care\n\nOsteoarthritis (OA) i
 s a chronic condition which affects over 300 million people globally and i
 s a leading cause of disability, yet predictive models often remain monomo
 dal, static, and opaque to clinicians. This dissertation develops OAAgent,
  a multimodal large language model (LLM) clinical assista...\n\n\nPegah Ah
 adian (Kent State University)\n---------------------\nLeveraging Performan
 ce Portability for High-Fidelity Simulations of Black Hole Accretion\n\nAc
 curately interpreting observations from the Event Horizon Telescope (EHT) 
 requires general relativistic magnetohydrodynamics (GRMHD) simulations tha
 t can model increasingly complex physics. To address this need within an e
 volving and heterogeneous computing landscape, we present some results fro
 m ...\n\n\nVedant Dhruv (University of Illinois Urbana-Champaign)\n-------
 --------------\nTowards Predictive Digital Twins with Applications to Prec
 ision Oncology\n\nWell calibrated mathematical and computational models en
 able the prediction and control of complex systems. These models can be ut
 ilized to design engineering systems or to develop treatment protocols. In
  contrast to one-size-fits-all approaches that seek to mitigate risk at th
 e population level, di...\n\n\nGraham Pash (The University of Texas at Aus
 tin, Oden Institute for Computational Engineering and Sciences)\n---------
 ------------\nStructural Equation Modeling for Heterogeneous Platforms\n\n
 Heterogeneous platforms introduce new complexities into performance modeli
 ng and prediction. The intrinsic performance asymmetries found in these pl
 atforms require radically different approaches to manage the compute diver
 sity of this polymorphic architectural design space. Core heterogeneity au
 gmen...\n\n\nSteven Harris (Washington University in St. Louis)\n---------
 ------------\nAI-Driven Resource Optimization for High Performance Computi
 ng: A Comprehensive Framework\n\nShared HPC centers are often underutilize
 d because jobs are commonly mis-specified for walltime, memory, and accele
 rators. This mis-specification causes queue churn, idle hardware, and long
  turnaround times. The main challenge is structural: researchers face a st
 eep learning curve across different n...\n\n\nManikya Swathi Vallabhajosyu
 la (The Ohio State University)\n---------------------\nTaming the Beast of
  Dynamic Resource Management in HPC\n\nDynamic resource management (DRM) e
 nables the resources assigned to a job to be adjusted during execution. Fr
 om a system perspective, DRM adds flexibility to resource allocation and j
 ob scheduling, with the potential to improve utilization, throughput, ener
 gy efficiency, and responsiveness. From an ...\n\n\nDominik Huber (Technic
 al University of Munich)\n---------------------\nAdvancing Data Center Wor
 kloads with Data Processing Units\n\nModern data center workloads demand s
 ubstantial server resources, motivating the adoption of data processing un
 its (DPUs) for improved efficiency. Despite increasing deployment, systema
 tic characterization of SoC-based DPUs remains limited. We present a rigor
 ous evaluation of NVIDIA’s BlueFiel...\n\n\nArjun Kashyap (University of C
 alifornia, Merced)\n---------------------\nA Framework for Digital Twins o
 f Future Quantum Clouds\n\nQuantum computing has emerged as a transformati
 ve technology capable of solving complex problems beyond classical systems
 ' limits. However, present-day quantum systems face critical bottlenecks, 
 including limited qubit counts, brief coherence intervals, and high error 
 susceptibility, which obstruct ...\n\n\nWaylon Luo (Kent State University)
 \n---------------------\nAccelerating Sparse Tensor Contractions\n\nSparse
  tensor contractions (SpTC) are a bottleneck for several algorithms in sci
 entific computing, data science, artificial intelligence and graphics. The
  SpTC operation is any expression of the form R(l0,l1, r0) = X(l0, l1, c0)
  * Y(r0, c0) where two tensors are multiplied along several dimensions t..
 .\n\n\nSaurabh Raje (University of Utah)\n---------------------\nManaging 
 Heterogeneous Topologies and Understanding Their Impact on Performance\n\n
 To solve increasingly complex problems more efficiently, modern HPC system
 s feature highly heterogeneous components: CPUs, GPUs, and recently QPUs (
 quantum processing units), each with a unique, complex compute topology. T
 he massive parallelism of GPUs, combined with emerging memory technologies
  on ...\n\n\nStepan Vanecek (Technical University of Munich)\n\nTag: Resea
 rch & ACM SRC Posters\n\nRecording: Not Livestreamed, Not Recorded\n\nRegi
 stration Category: Technical Program Reg Pass\n\nSession Chairs: Kento Sat
 o (RIKEN Center for Computational Science (R-CCS)); Chris Schlipalius (Paw
 sey Supercomputing Research Centre; Commonwealth Scientific and Industrial
  Research Organisation (CSIRO), Australia); and Anja Gerbes (Georg-August-
 Universität Göttingen)
END:VEVENT
END:VCALENDAR