Philip Bachman

amdim-public. Public repo for Augmented Multiscale Deep InfoMax representation learning

NN-Python. The useful and used parts of NN-Dropout

Pseudo-Ensembles. Materials for implementing and reproducing results in the NIPS paper.

Sequential-Generation. Generative models and other stuff too, maybe, perhaps even probably

NN-Dropout. This code is for training feedforward neural networks with dropout.

MatNets-NIPS. Repo for code for the NIPS paper entitled "An Architecture for Deep, Hierarchical Generative Models"

ICML-2015. Code for models in my ICML 2015 paper on VAEs, GSNs, etc.

MultiClassBoosting. Code for a variety of approaches to learning classifiers via boosting

Xcetera. Assorted code fragments

GradReg. Code for ICML 2014 paper: "Sample-based Approximate Regularization"

Matrix-Learning. Code for algorithms and tests pertaining to sparse network basis learning