Complexity beyond parameter counting - SLT Seminar 54

Dan Murfet talks through some preliminary work (SLT40) exploring how run-time error-correction can contribute to the local learning coefficient of a Turing machine parametrised by a UTM, in the proper context.
This talk is from the Singular Learning Theory seminar: metauni.org/slt
About: Singularities are knowledge. A learning seminar on Watanabe’s Singular Learning Theory: algebraic geometry serves statistical learning theory.
This video was recorded on 14/12/23, as a part of metauni day 40.
You can join this seminar from anywhere, on any device, at www.metauni.org. All are welcome.
#metauni #slt