Quantitative Aesthetics : Introduction to Machine Learning for Design

This course aims to introduce students in art and design related fields to concepts and techniques from Machine Learning. Rather than focusing on large “black box” generative AI models we are going to explore the design potential and implications of some of their constituent components and the frameworks that enable them. Through a series of workshops and small projects student should develop an intuitive understanding of how model architecture, training and inference work, how to explore and visualize embeddings and latent spaces and how text and images can occupy the same semantic space. The emphasis will be placed more on the perceptual capabilities and idiosyncrasies of ML models with some forays into generative processes.

The course will start with a practical introduction to two- and three-dimensional vector math as this will be the basis for extending our intuition to the multidimensional vector spaces at the heart of ML models. In parallel we will be learning the fundamentals of python programming.

Our main tool chain will consist of Python with the ML library Pytorch and the Computer Vision Library OpenCV along with Rhino’s grasshopper visual programming environment that will help us visualize vector spaces and handle parameterized geometric inputs.

The first day of classes, Tuesday, September 3rd, is held as a MONDAY schedule at the GSD. As this course meets on Monday, the first meeting of this course will be on Tuesday, September 3rd. It will meet regularly thereafter.