#### Learning the nature of dark energy from weak gravitational lensing

We would like to compute Bayesian credible regions for certain parameters of cosmological models of the universe given data from weak gravitational lensing.

This will involve developing a workflow for applying dimensionality reduction and statistical modeling to cosmological simulation data, and (eventually) data from the Canada France Hawaii Lensing Survey.

Requirements: machine learning, statistical inference, experience with scipy/numpy/scikit-learn.

#### Computing exploration policies for contextual bandit problems

We would like to develop and experimentally evaluate an efficient implementation of a recently proposed algorithm for computing exploration policies for use in contextual bandit settings.

This will involve coding an efficient implementation of an abstract coordinate descent algorithm that makes calls to a supervised learning algorithm through a generic interface, so that it is possible to plug-in existing learning algorithms for models like decision trees and neural networks.

Requirements: machine learning, solid programming skills, experience with machine learning toolkits.

#### Forecasting influenza-like illnesses using search query rates

We would like to develop methods for forecasting the prevalence of influenza-like illnesses using Google search query rates, much like the recently-defunct Google Flu Trends.

The first step will be to implement the method most recently used by Google using open-source machine learning toolkits.

Requirements: machine learning, statistical inference, experience with scipy/numpy/scikit-learn.

##### Interested? E-mail me and we can talk.