I was part of the Technarte 2022 lineup with a talk on small datasets and their creative potential.
I talked about my preference for hand picked datasets and how one can study the logics and limitations of deep learning models with them. This is a starting point for reimagining and rebuilding machine learning practices that currently tend to be very opaque.
The talk was inspired by Dan McQuillan’s ideas on 'machine learning for the people' and Tega Brain’s idea of eccentric engineering.
