What are you modeling?

In this post, we will explore how Discriminative/Generative and Frequentist/Bayesian algorithms make different decisions about what variables to model probabilistically. There are many ways to characterize Machine Learning algorithms. This is a direct consequence of the rich history, broad applicability and interdisciplinary nature of the field. One of the clearest... [Read More]
Tags: Machine Learning, Probability, Discriminative, Generative, Frequentist, Bayesian

Compositional Structures in Machine Learning

As researchers apply Machine Learning to increasingly complex tasks, there is mounting interest in strategies for combining multiple simple models into more powerful algorithms. In this post we will explore some of these techniques. We will use a little bit of language from Category Theory, but not much. In the... [Read More]
Tags: Machine Learning, Neural Networks, Category Theory, Composition

Some Thoughts on ICLR 2019

I recently attended ICLR 2019 in New Orleans, and I was lucky to have the opportunity to show off our paper on a novel attention module and image understanding dataset. I really enjoyed the entire conference, and I thought I’d share brief overviews of two of my favorite presentations from... [Read More]
Tags: ICLR, Machine Learning, ML, Neural Network, Conference

My solutions to Bartosz Milewski's "Category Theory for Programmers"

I recently worked through Bartosz Milewski’s excellent free book “Category Theory for Programmers.” The book is available online here and here. I had an awesome time reading the book and learning about Category Theory so I figured I’d post my solutions to the book problems online to make it easier... [Read More]
Tags: Category Theory, Functional Programming, Mathematics, Solutions

Introducing repcomp - A Python Package for Comparing Trained Embedding Models

When I’m building models, I frequently run into situations where I’ve trained multiple models over a few datasets or tasks and I’m curious about how they compare. For instance, it’s clear that if I train two word vector models on random subsets of Wikipedia, the trained models will be “similar”... [Read More]
Tags: Embeddings, Machine Learning, ML, Python, Comparison, Neural Network, Word Vector