The Price of Imbalance

May 20, 2026    
The Price of Imbalance

When evaluating a machine learning model, most practitioners simply apply the standard precision/recall formulas to whatever test set they have. The results can be wildly (Read more...)

Adversarial Neural Cryptography

Oct 20, 2018    
Adversarial Neural Cryptography

In 2016, researchers from Google Brain published a paper showing how neural networks can learn symmetric encryption to protect information from other AI attackers. (Read more...)

Visualizing CVAE

Oct 12, 2018    
Visualizing CVAE

The encoding vector of a Conditional Variational Autoencoder (CVAE) is comprised of two components - the style encoding, and the conditioning vector. This visualization explores (Read more...)

Variational Autoencoders

Sep 10, 2018    
Variational Autoencoders

We saw in Intro to Autoencoders how the encoding space was a non-convex manifold, which makes basic autoencoders a poor choice for generative models. Variational autoencoders fix this issue by ensuring (Read more...)

Intro to Autoencoders

Aug 29, 2018    
Intro to Autoencoders

Solving problems using machine learning often involves looking at data with thousands of features. It is often possible to reduce the number of features, which speeds up training and is also useful for data visualization. One sophisiticated way to do so is using autoencoders (Read more...)