Introduction to Data Analysis and Machine Learning in Physics
Overview
Data analysis belongs to the core elements of physics. At the same time
machine learning is extremely successful in fields like image recognition,
speech processing and medical diagnosis. In physics machine learning is
becoming increasingly important. This course gives a
practical introduction to both topics.
- lecturers: Martino Borsato, Jörg Marks, Klaus Reygers
- The course is offered as a virtual event usinf heiCONF
- Date: 6.4. - 9.4.20201 9:00-12:00 Uhr and 14:00-17:00 Uhr
Contents
- Fitting models to data
- Overview: Machine Learning
- Linear models, logistic regression
- Boosted Decision Trees
- Neural networks
Programming language
We'll use python 3 in the course. Basic knowdegle of the language is required
for this course. We'll work a lot with
jupyter
notebooks.
A nice summary of
important python commands
is available on the website of the Stanford lecture CS231n.
For machine learning applications we'll use the libraries
scikit-learn
and
Keras.
Target group
Students with Python knowledge.
Maximum number of participants
50 corresponding the size of the exercise goups. A course registration is need
and starts the 16th of March (10am)
Participation requirements
In order to participate you need a computer/laptop with on operating system of
you choice with a current web browser and a user-id in the CIP pool of the
faculty of physics. We will use the jupyter hub of the CIP pool
jupyter2.kip.uni-heidelberg.de.
In addition for communication a camera,
microphone and speakers are needed. We will use heiCONF as communication
platform with several breakout rooms for execises in small groups.
In the document
courseInformation.pdf
the heiCONF links and technical hints can be found. We would like to ask you
to meet the technical requirements in advance of the beginning of the course.