Deep Learning Workshop

October 30, 2017
NCSA Building Auditorium
1205 W. Clark St., Urbana, Illinois

NCSA will soon be installing a powerful system for deep-learning as part of a major research instrumentation (MRI) grant from the National Science Foundation. The purpose of this award is to deploy a multi-node multi-GPU system and develop tools, techniques, and algorithms to better support deep learning applications on this system. This workshop serves three goals:

  1. Introduce the technology used in this system
  2. Invite faculty and research groups to make use of this system to support both the development work for the MRI project and their research and describe the timeline for availability
  3. Begin to build a community of researchers around deep learning, with the MRI project an initial focus

The workshop will start with a presentation by IBM about PowerAI, which will be available on the proposed deep learning infrastructure. We will then discuss research opportunities in scalable deep learning and the infrastructure to be developed in collaboration with IBM and NVIDIA.

Please fill out the registration form if you intend to attend any part of the workshop. Also please fill out the survey which will help us to better understand current trends and needs of the University of Illinois deep learning community.

Agenda

  • 10:00 a.m. – noon — IBM PowerAI Introduction
  • 2:00–3:00 p.m. — Lighting Talks
    • "Deep Learning for Multimessenger Astronomy" — Eliu Huerta Escudero
    • "Using a Single Deep Network for Multiple Tasks" — Arun Mallya
    • "Stochastic Variational Video Prediction" — Mohammad Babaeizadeh
    • "Entity Based Visual Scene Understanding" — Christopher Cervantes
    • "Deep Learning at NCSA Genomics" — Luda Mainzer
    • "Learning Collective Variables of Molecular Structures with Hierarchical Data-Augmented Autoencoders" — Wei Chen
    • "Uncertainty Estimation in Deep Neural Networks and Applications to Active and Multi-Task Learning" — Shubhanshu Mishra
    • "Deep Learning for Solving High-Dimensional Partial Differential Equations" — Justin Sirignano
    • "Deep Learning for an Automated Pipeline from the MR Scanner to the 3D Printer for the Heart" — Xi (Steve) Peng
    • "CarML" — Abdul Dakkak
    • "Deep Learning for Time-Series Signal Processing Applied for LIGO Gravitational Wave Analysis" — Daniel George
    • "Computer Vision with Deep Learning" — David Forsyth
  • 3:00–4:00 p.m. — NSF MRI Project Introduction and Discussion
    • "Community Survey Results" — Roy Campbell
    • "MRI: Development of an Instrument for Deep Learning Research" — William Gropp
    • "MRI System" — Volodymyr Kindratenko
  • 4:00–5:00 p.m. — Research Panel
    • Roy Campbell (panel chair)
    • Wen-Mei Hwu, Prof. ECE
    • Jian Peng, Assist. Prof. CS, project co-PI
    • Paris Smaragdis, Assoc. Prof. ECE/CS
    • Linton Ward, IBM Distinguished Engineer, Cognitive Systems Solutions