From Bias to Knowledge: The Epistemology of Machine Learning (2023 - 2029)

Abstract

There are various biases which enter into the development of a machine learning system. Relatively unexplored so far is the concept of inductive bias, which lies at the very heart of machine learning: inductive biases are the assumptions that allow a learning algorithm to learn. This project connects the philosophy of science and the mathematical theory of machine learning to clarify the concept of inductive bias. This better understanding of inductive bias will be the central element in a general epistemological account of how we gain knowledge through machine learning methods.

Project information

Project title
From Bias to Knowledge: The Epistemology of Machine Learning
Funded by
DFG Emmy Noether Programme
Project link
-
Project duration
2023 - 2029
Funds awarded
-
Project team
Dr. Tom Sterkenburg (principal investigator)
Associated Chair
Chair of Philosophy of Science