A seminar at KISD on understanding (deep) machine learning through building and using machine learning systems.
Over the past few years, a renaissance of machine learning took place. Fueled by a new abundance of affordable computing power as well as cloud-based code-sharing, “deep” neural networks and similar methods have fundamentally changed how artificial intelligence is understood and applied. This development highlights and radicalizes problems that are inherent to computing (such as bias), but also introduces novel problems (such as the difficulty to obtain causal explanations from a neural network). As these methods are becoming part of the interfaces we use every day, it is time, for us as interaction designers, to obtain an understanding of how machine learning actually works.
In this seminar, we will build deep learning systems in a bottom-up manner, to understand how they are created and how they work. We will complement this with a top-down look on machine learning products and APIs that already exist. And we will try to get an understanding of the area by looking at the history and theory the field.