- Start
- End
- Types of event
- Vortrag
- Venue
-
Ernst-Abbe-Platz 8, HS028
07743 Jena
Google Maps site planExternal link - Language of the event
- English
- Barrier-free access
- No
- Public
- Yes
Analyzing Neural Network Decisions and Detecting Biases (Niklas Penzel)
With a short introductional lecture by Maya Köthe (research assistance/ Gender in Focus): “Gender Bias Mitigation in Artificial Intelligence“
dunkler Hintergrund neuronale Netzwerke
Image: Adobe Stock, KI-Inhalt
Event details
About the lecture (in English)
Deep learning models achieve high predictive performance but lack intrinsic interpretability, hindering our understanding of the learned prediction behavior. Because they operate as 'black boxes,' these systems risk learning spurious correlations. Thus, they might amplify unintended, harmful biases contained within the data. Explainable AI (XAI) provides the diagnostic tools we need to audit the model's decision-making process, and systematically uncover these biases. In this talk, we will discuss how neural networks learn biases and how this can be investigated. In part, we will focus our discussion on gender in medical image analysis.
Before the lecture by Niklas Penzel the research assistent of "Gender in Focus" Maya Köthe will give a short introductional lecture on “Gender Bias Mitigation in Artificial Intelligence“.