Ontological Surprises: A Relational Perspective on Machine Learning
Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review
This paper investigates how we might rethink design as the technological crafting of human-machine relations in the context of a machine learning technique called neural networks. It analyzes Google’s Inceptionism project, which uses neural networks for image recognition. The surprising output of one the experiments reveals that such networks might be used to trace relations between entities. This paper contributes by fleshing out the necessary changes in the ways HCI builds, tests, and engages neural networks in the design of interactive systems from a relational perspective; it proposes a hybrid approach where machine learning algorithms are used to identify objects as well as connections between them; finally, it argues for remaining open to ontological surprises in machine learning as they may enable the crafting of different relations with and through technologies.
|Title of host publication||Proceedings of the 2016 ACM Conference on Designing Interactive Systems|
|Number of pages||5|
|Publisher||Association for Computing Machinery|
|Publication status||Published - 2016|