2019-04-11 Intel Artificial Intelligence Workshop (HIAI1S19)

Date:Thursday, April 11, 2019, 9:00-17:00
Location:LRZ Building, Garching/Munich, Boltzmannstr. 1, LRZ Hörsaal
Contents:

The workshop will highlight the usage of Intel's optimized performance libraries underneath popular machine learning frameworks. It includes hands-on sessions around classical machine learning as well as deep learning on HPC systems. Besides introducing the transparent usage of performance libraries, the focus will be to empower participants to scale their deep learning (TensorFlow) application training on distributed HPC systems. The hands-on examples are written in Python.


Agenda:

1) Introduction

2) Overview Machine Learning / Deep Learning

3) Classic Machine Learning Tools

1. Intel performance Libraries - MKL & DAAL
2. Intel Distribution for Python (IDP)
3. IDP Hands on labs
    - NumPy & MKL
    - K-Means Clustering & DAAL
    - SVM & DAAL

4) Deep Learning Tools

1. Intel performance Libraries - MKL-DNN & MLSL
2. Intel optimized Frameworks / TensorFlow
3. TensorFlow Image Classification hands-on
    - Simple Softmax
    - Simple CNN
    - Monitored Training Session
    - Distributed class. with Horovod
    - Custom Images and Batches
    - Benchmarking distributed TensorFlow


Prerequisites:

The course requires participants to have some prior experience with Python and SSH (if in doubt, for the latter please attend Introduction to the LRZ Supercomputing and Machine Learning Infrastructure).
All participants are expected to bring their own laptops and have an SSH client installed.

Language:English
Instructors:Michael Steyer, Fabio Baruffa (Intel), J. Albert-von der Gönna, D. Brayford (LRZ)
Registration:Via the LRZ registration form. Please choose course HIAI1S19.
Slides:

Download

If you attended this course, please don't forget to rate it. We highly appreciate your feedback:
https://survey.lrz.de/index.php/433414