Austrian Early Career Conference 2024

Contribution:
Talk

Authors:
I. Piantschitsch [1,2]

Affiliations:
1: Institute of Physics, University of Graz, Austria; 2: Department of Physics, University of the Balearic Islands (UIB), Spain

Title:
Deep Learning & Philosophy - On the epistemic role of Deep Learning in science

Abstract:
Deep learning networks are ubiquitous in many branches of science and engineering, however, not all the details of their underlying mathematical structure is comprehensively understood. This fact undoubtedly raises questions regarding the use of deep learning as a valid scientific method but also regarding its relation, its differences, and its similarities to computer simulations, experiments, and observations. The interdisciplinary project "DeLPhi – Deep Learning & Philosophy", which is funded by the Styrian government, aims to analyse benefits, challenges, and risks of applying deep learning networks in science. The international team working on this analysis includes experts in the fields of mathematics, physics, cognitive science, philosophy, law, and art. The objective of the project is the mutual exchange of expertise in the different fields, such as the mathematical background of deep learning, its application in astrophysics, its epistemic content but also its implications on technology law for instance.