Weitere Veranstaltungen in Regensburg

AI, federated learning and digital platforms at the interface of chances, trust and social change

Tickets an der Abendkasse erhältlich
Mi 20 Mai Doors open at 18:30 - Event starts at 19:00 - Come by and have a beer and some food before we start!
Murphy's Law, Spiegelgasse 6
93047, Regensburg
Tickets Preis Mng.
Standard €2.50
Spende Pint of Science helfen

Verbleibende Tickets: 40

AI and Federation have had a profound impact on how we live, work and interact.

AI has changed how we approach problem solving as groups, while federation promissed to protect us from tech conglomerates influence in all aspects of life. 

Find out more at Wednesdays event of Pint of Science in Regensburg! 

Trustworthy Federated Learning

Rawad Bitar (Researcher)
As the AI revolution takes shape, the need for reliable and extensive training data becomes more and more important. Federated learning (FL), a framework introduced by Google researchers, allows training neural networks on users' devices without moving their data to a central server; thus hitting two birds with one stone: On one hand the users data remains local, giving a sense of privacy, and on the other hand the burden of communicating and storing this data is circumvented.
However, is privacy really guaranteed? Furthermore, FL relies on trusting all participating clients, but what happens if some of them deviate from the protocol?
...

Wie generative KI die kollektive Problemlösung neu gestaltet

Robin Nowak (Doctoral Researcher)
Diese Studie untersucht, wie Fortschritte im Bereich der generativen Künstlichen Intelligenz (KI) die Relevanz menschlicher Gruppen für die Problemlösung verändern. Mit zunehmenden KI-Fähigkeiten verschiebt sich die Problemlösung immer mehr vom Crowdsourcing-basierten Austausch hin zu eins-zu-eins-Interaktionen mit der KI.

Das Ausmaß dieser Substitution – und damit die Frage, wo menschliche Gruppen für die Problemlösung weiterhin relevant bleiben – hängt von zwei Hauptfaktoren ab: der Verfügbarkeit von Vorwissen und der inhärenten Komplexität eines Problems.

...
© die Mitwirkenden OpenStreetMap