I recently read Ferenc Huszár's blog post summarizing recent research on how evolutionary strategies can be used in place of gradient descent or back-propagation. That research focuses on how evolutionary strategies for reinforcement learning get the same results as back-propagation, but the computation is more parallelizable and therefore faster ...

Continue reading

I've grown increasingly frustrated with the way the popular press talks about Machine Learning and Artificial Intelligence.

Too many articles talk about Machine Learning (ML) as a sci-fi black box capable of doing and learning anything, and leave the reader with a poorer understanding of what ML is than ...

Continue reading

A self-lead refresher course in basic ML algorithms

I'm in the process of implementing various machine learning algorithms from scratch.

For now the algorithms include:

• Regression (logistic and least squares) via gradient descent

• Decision Trees

• Random Forests

I'll be benchmarking these algorithms on the handwritten digits dataset that ...

Continue reading