Peter Leonard

DH-seminariet: When algorithms dream: Neural networks, latent space, and cultural heritage

Kultur & språk
Naturvetenskap & IT

This talk considers the possibilities of different kinds of artificial neural networks for the digital humanities: networks that enhance interpretive ability, and networks that can be sampled to generate novel forms of expression. We consider both textual and visual datasets, with a focus on Nordic cultural heritage, and consider neural networks for their ability to produce latent spaces: theoretical realms of cultural production that spring from – but are not bounded by – the historical record. As more literary and visual collections are digitized and placed online, it seems fair to consider both the work and play of big data: what happens, in other words, when algorithms dream.

3 dec 2021
15:15 - 17:00
J233, Humanisten, Renströmsgatan 6, Göteborg
Ytterligare information
Seminariet streamas även via

Bra att veta
Deltagare på plats välkomnas till ett postseminarium.
Peter Leonard


• Melvin Wevers and Thomas Smits, “The Visual Digital Turn: Using Neural Networks to Study Historical Images” Digital Scholarship in the Humanities, Volume 35, Issue 1, April 2020.

• Kashmir Hill and Jeremy White, “Designed to Deceive: Do These People Look Real to You?” The New York Times, November 21, 2020

Peter Leonard is the Director of the Digital Humanities Lab, Lecturer in Data Science, and Fellow at the Whitney Humanities Center at Yale University.  He was a Fulbright Fellow at Uppsala University during 2007-2008, and completed his PhD in Scandinavian Literature at the University of Washington in 2011.