From a computer’s standpoint, the difficulty in pattern recognition is one of surplus: with an endless variety of patterns, all… (more…)

Getting creative with neural networks is easier than you think.

Michael Kearns, University of Pennsylvania Algorithmic Game Theory and Practice https://simons.berkeley.edu/talks/michael-kearns-2015-11-19...

The Oscars and their preferential balloting led me to create a novel machine learning approach to mimic this voting system... (more…)

My experiments with Spleeter, a jaw-dropping new library that quickly and freely separates vocals from music with machine learning. (more…)