(Un)supervised (Con)fabulation - Machine Learning and Time
Project information
submitted by
Christina Schinzel
Mentors
Prof. Ursula Damm
Faculty:
Art and Design
Degree programme:
Media Art and Design -Study programme Media Art und Design (MAD) (english) (Master of Fine Arts (M.F.A.))
Type of project presentation
Final project
Semester
Summer semester 2022
- Marienstraße 7b
(201) - Marienstraße 7b
(201)
Project description
What is Machine Learning? What is time? And how are the two related? These are the principal questions of this work. It is about temporal structures that exist inside and are created by machine learning systems, where past data serves as a basis for future decisions.
In this work the same principle is applied to videos and is realized in the form of a video installation. Starting with real videos, that are recalculated by a machine learning algorithm, by which the next image on the videographic timeline is generated on the basis of the previous images. In the end the original video gradually disppears behind the algorithmic calculations and can only be sensed in fragmented abstraction.
The objective of this work is a visualization of the invisible: of the ephemerality of time and the invisible logical level of computer code of machine learning, which captures, processes, calculates, transforms and also repeats temporal events in the form of data. This effect can be described as "the enduring ephemeral" (Wendy Hui Kyong Chun: Programmed Visions). This video installation attempts to capture this "enduring ephemeral" of machine learning.
The project was funded by the summaeryFonds of University Communications 2022.