Why do some machine learning models fail?


Most of the Machine Learning talks present beautiful cases of success, but in reality ml models often fail to deliver the desired performance. It is not uncommon to see developers blaming certain ml models and even providing blacklists of ml models. In this talk, I will provide some tips on choosing ml models and guide them through the path of finding a good solution. I will also present two of my recent works that use machine learning in astrophysics and in neuroscience.

Language: English

Level: Beginner

Rafael Garcia-Dias

Research associate - King's College London

Postdoc associate researcher at the King's College London. Working at KCL with the Machine Learning in Mental Health project. Currently, my main focus is to develop machine learning models to diagnose patients based on structural MRI. In general, my main interest is to benefit from the huge amount of data being acquired in neuroscience, using machine learning, to make a positive impact on the field.

Go to speaker's detail