Anatomies of Intelligence
Symposium:
Title:
- Anatomies of Intelligence
Organiser/Presenter(s):
Statement:
The research conducted under the Anatomies of intelligence project has been feeding a growing dataset and online repository which gathers terminologies and techniques for a critical examination of the “anatomy” of learning and prediction processes. The same platform is used to explore, through performance, how such a collection and an artisanal algorithmic toolkit can confront the idealized bodies of artificial intelligence — its representational structures and sense-making processes.
Participants will explore together with us the concepts of the project, such as connections between early Enlightenment anatomical science, classification and collection and contemporary practices in data collection and statistical machine learning. We also lead the participants through embodiment and inner-movement practices taken from dance, and guide them through exercises which connect their own sense of body and internal orientations to the high-dimensional spaces and complex topologies of feature vectors within machine learning.
The workshop culminates with a collection and clustering exercise, we will focus specifically on embodying the widely used unsupervised clustering algorithm K-means and collectively deconstruct its underlying assumptions of (Euclidean) space and distance. Core to the workshop will be a consideration of data classification and clustering. What different kinds of sense-making emerge when handling of data is grounded in the experience of having a body? What is a unit of anatomical space and how can that inform embodied approaches to machine learning?
Whereby participants will experience in a “hands-on” manner the process of optimization inherent in statistical learning, while also becoming acutely aware of the multiple vectors where subjective bias is entangled within this process.
