Supplementary material for the BCTP Lecture Series
On this Github repository, I collect files I coded during the afternoon sessions of the BCTP Lecture Series Machine Learning and its Applications. Note that all examples can be drastically improved. The point is not to find the best neural network, but to illustrate the concepts we have discussed in class.
Monday
In this session, we created our first feed-forward neural network that learns the xor function. We implemented the neural network in Mathematica, Keras (with tensorflow backend), and PyTorch. The files are available here.
Tuesday
In this session, we created our first convolutional neural network using PyTorch. We classify galaxies into spiral, elliptical or unknown. The data is provided by the Galaxy Zoo project. See http://adsabs.harvard.edu/abs/2008MNRAS.389.1179L for more details. The pictures of the galaxies themselves are provided by the Sloane Digital Sky Survey. The files are available here.
NN predicting this is a spiral galaxy:
(Image source: https://www.sdss.org)
Wednesday
In this session, we demonstrated how to code an environment that can be connected via the OpenAI gym to ChainerRL. We illustrate how the the A3C agent finds good energy configurations for the 1D Ising model by flipping spins at any of the lattice sites. The files are available here.
I found an optimal configuration!
↑ ↑ ↑ ↑ ↑ ↑ ↑ ↑ ↑ ↑ ↑ ↑ ↑ ↑ ↑
I started from
↑ ↑ ↓ ↓ ↑ ↑ ↑ ↑ ↑ ↓ ↓ ↑ ↓ ↑ ↑
and took the actions
[10, 2, 3, 9, 12]
Thursday
In this session, we illustrated different unsupervised clustering algorithms (k-means, mean shift, DBSCAN, Birch) that were discussed in class using scikit learn. The files are available here.
