If you are reading this, I am almost certain you have heard the term Deep Learning and/or Pokemon. The former is a branch of Machine Learning that has risen in popularity in the recent years, and the latter is a franchise that has been popular since the 90’s and with the release of Pokemon Go everyone has been talking about it.

What is the connection with these two concepts? Out of curiosity I took on the task of attempting to predict the type or category of a given Pokemon using its appearance, in particular its color, using a Deep Learning algorithm called Convolutional Neural Network or ConvNet.

This experiment was performed using Python, the deep learning library TensorFlow version 0.10.0 and its simplified interface TF Learn.

In this repository you can find the script used for the experiment.

The full report is available at: Pokemon, Colors and Deep Learning