December 2019
Organizers: Heinrich Matzinger & Greg Mayer
Registration
Registration is free. Please email Dr. Matzinger, matzi@math.gatech.edu to receive a preparation package for the mandatory preparation exercises in Python (about 3 hours).
Abstract
The workshop contains talks on results from high-dimensional statistics and machine learning which are relevant to practitioners. It also contains a mini Machine Learning crash course on Thursday and Friday, based on real data intuition and mathematics. Crash course sessions are mixed with hands on programming with Python, Numpy, Pytorch, Fastai. Those already familiar with Machine Learning can join us on Saturday and skip the Thursday and Friday sessions.
Schedule
Thur Dec 5 – Skiles 156
- 1500-1600 Welcome and quick intro into Basic classification. Greg Mayer. ClassificationIntro.pdf.
- 1600-1700 Introduction to Python programming with Anaconda. Greg Mayer. PythonWithAnaconda.pdf and CrashCourse2019.ipynb.zip
- 1700-1800 First part of a hands-on mini project: text classification comparing methods using NumPy. Heinrich Matzinger and Greg Mayer. TextClassification.pdf
Fri Dec 6 – Skiles 257
Files for hands-on programming:
EM Slides: EM.pdf
- 1100-1130 Welcome and introduction.Heinrich Matzinger and Greg Mayer
- 1130-1230 Numpy programming hands on. Heinrich Matzinger and Greg Mayer
- 1230-1300 Lunch
- 13-14 Intro to intuitive regression, EM, Principal Components, text classification. Heinrich Matzinger
- 14-16 Hands on project programming. Heinrich Matzinger and Greg Mayer
- Hidden Markov Models Part 1: ett-Atlanta19
Sat Dec 7 – Skiles 268
- 1100-1200 Data compression with random matrices. Konstantin Tikomirov, Georgia Tech
- 1200-1230 Lunch
- 1230-1330 Hidden Markov Models. Jurl Lember, Tartu University, Estonia
- 1330-1430 Exploring low-dimensional structures in data science. Wenjin Liao
- 1430-1530 Distributed SVD. Raphael Hauser, Oxford University
- 1530-1630 The differential equation method: dynamic concentration Warnke Lutz, Georgia Tech
- 1630-1900 Deep Learning with fast.ai and/or Programming Deep net with fastai. Heinrich Matzinger
Sun Dec 8 – Skiles 268
- 1100-1200 Deep learning with fastai, part II and programming. Heinrich Matzinger & Greg Mayer
- 1200-1230 Lunch
- 1230-1500 TBD