AI Colloquium: Recognition of Street Scenes using Neural Networks
The AI Colloquium is a lecture series at the TU Dortmund University, a partner institution of the Lamarr Institute, presenting groundbreaking research in the field of Machine Learning and Artificial Intelligence. The upcoming colloquium, on April 18th from 10:15 AM to 11:45 AM, will focus on the topic “Uncertainty Quantification and Robustness in Recognition of Street Scenes using Neural Networks,” presented by Prof. Dr. Hanno Gottschalk from the TU Berlin University and organized by our scientific coordinator Dr. Jens Buß.
Prof. Gottschalk, a renowned expert in mathematical modeling of industrial life cycles, will address the challenges neural networks face in recognizing street scenes, particularly when encountering Out-of-Distribution (OoD) data. Furthermore, he will explore methods to identify and localize these OoD objects and discuss strategies to enhance network robustness against domain shifts using pre-training with image-text combinations.
The 90-minute session will consist of a one-hour lecture in English followed by a 30-minute Q&A session. Whether you’re an experienced researcher or an AI enthusiast, we look forward to your attendance and active participation.
Details
Date
18. - 18. April 2024
Location
Joseph-von-Fraunhofer Strasse 25, Room 303 + Zoom