-
Notifications
You must be signed in to change notification settings - Fork 1
Home
AstroHack Week is a week-long summer school / hack week / unconference focused on astrostatistics and data-intensive astronomy. The vision is to provide a format to encourage collaboration and sharing of expertise, so that both young and experienced astronomical researchers will benefit from the week.
Each morning, we will spend 2-3 hours in a typical "summer school" format, focused on introductory statistics, data mining, and machine learning as they pertain to astronomical research. The remainder of the day will be spent in an "unconference" format: there will be space provided for working solo or with other people, for having breakout sessions on various topics, and for collaborating to work on particular problems, data sets, or research approaches. This will not be a week away from serious research, but a week spent on focused research time within an environment where new ideas and approaches can be collaboratively developed.
It will take place at the University of Washington from September 15-19, 2014. The hope is that the Moore/Sloan Data Science Institute space will be completed in time; if not, we have the Active Learning Classroom space reserved as a backup venue.
Here are the folks we've contacted who will be involved with planning & teaching (this list may grow or shrink with time).
- Jake Vanderplas: Director of Research in Physical Sciences, University of Washington eScience Institute
- Zeljko Ivezic: Professor, University of Washington Astronomy Dept. & Project Scientist, LSST
- David W. Hogg: Professor, New York University Physics Dept. & Visiting Professor, Max Planck Institute for Astronomy
- Phil Marshall: Staff Scientist, Kavli Institute for Particle Astrophysics and Cosmology, SLAC National Laboratory
- Berkeley representative? Josh or Fernando?
In addition, we have space for about 35 participants. The hope is that these participants would be drawn from a wide and diverse swath of academia: from young graduate students all the way up through postdocs, researchers, and faculty.
This conference is sponsored by University of Washington's eScience Institute, with support from the Gordon & Betty Moore Foundation and the Alfred P. Sloan Foundation.
Due to this sponsorship, attendees will only be required to pay a small fee (probably in the range of $50 - $70).
-
9:00am - noon : Intro to Data Analysis with Python
- Interactive Computing with IPython
- Effective Computing with NumPy
- Visualization with Matplotlib
- Exploring computational tools available in SciPy
-
1:00pm - 5:30pm : Hack time & Breakouts
-
5:30pm - 6:00pm : Daily Wrap-up
- 9:00am - noon : Introduction to Classical Statistics
- Intro to classical probability theory
- Maximum likelihood Optimization & Uncertainty Quantification
- Goodness of Fit and Hypothesis Testing
- Confidence Estimates using Bootstrap
- 1:00pm - 5:30pm : Hack time & Breakouts
- 5:30pm - 6:00pm : Daily Wrap-up
-
9:00am - noon : Introduction to Bayesian Statistics
- Bayes' Theorem and Bayesian probability
- Bayesian Priors
- Posterior optimization, marginalization, and Uncertainty Quantification
- Hypothesis Testing
- Brief intro to Markov Chain Monte Carlo (MCMC) sampling
-
1:00pm - 5:30pm : Hack time & Breakouts
-
5:30 - 6:00pm : Daily Wrap-up
-
9:00am - noon : Principles of Machine Learning: Supervised Learning
- Supervised Machine Learning: Classification vs Regression
- A survey of Classification techniques
- A survey of Regression techniques
-
1:00pm - 5:00pm : Hack time & Breakouts
-
6:00pm to Late : off-site dinner & hackathon
-
10:00am - noon : Principles of Machine Learning: Unsupervised Learning
- Clustering Algorithms
- Dimensionality Reduction Algorithms
- Density Estimation Algorithms
-
1:00 - 5:00 : Hack time & Breakouts
-
5:00 - 6:00 : Wrap-up