Markus Schedl gives a talk about Bias and Fairness in Recommender Systems
LIT Breakfast Briefing
Markus Schedl und Florian Poltschak
Our LIT Breakfast Briefings feature outstanding and impressive JKU researchers who talk to the scientific community about findings from their current research.
We present two topics and discussions in 30-minute sessions.
February 15, 2022, 8.30-9.00 a.m., online via zoom., öffnet eine externe URL in einem neuen Fenster
- Univ.-Prof. Markus Schedl (Institute of Computational Perception , öffnet eine externe URL in einem neuen Fenster/ LIT AI Lab, öffnet eine externe URL in einem neuen Fenster)
Bias and Fairness in Recommender Systems
Abstract: Recommender systems (RSs) are omnipresent. Their underlying machine learning algorithms influence which products we buy, which movies we watch, and which jobs we apply for. Unfortunately, these algorithms are prone to various biases, which can occur or intensify at different stages of the recommendation process. For instance, training data could be not representative of the population at large (data bias). Also, the adopted RS algorithm may amplify such imbalanced datasets (algorithmic bias).
These biases can cause unfair behavior of RSs, in which case they are not only harmful, but often also legally problematic. For instance, users with similar skills but different genders may receive job recommendations within different pay grades.
In this talk, I will introduce approaches to identify and mitigate data and algorithmic biases, in order to create recommendation technology that is fair, i.e., does not systematically discriminate against certain individuals or groups in favor of others.
- DI Dr. Florian Poltschak (JKU HOERBIGER Research Institute for Smart Actuators, öffnet eine externe URL in einem neuen Fenster)
How smart actuators can help solve societal challenges
The goals of industrial contract research are directly related to current social challenges. After all, it is often a matter of researching the foundations for new products that are to prove profitable on the market. The resulting research goals are essentially driven by innovation and profitability. In addition, especially in the industrial sector, political measures, for example in connection with the energy transition and sustainability, are also gaining in importance.Using selected examples, I will show how far this influences the objectives of current mechatronic research in the field of smart actuators and what contribution they can make.
The LIT Breakfast Briefing will be held in English.
Zoom: https://jku.zoom.us/j/92664331002?pwd=S1hLQTJ5cmZhMy8vMFJTejlwbFFjUT09, öffnet eine externe URL in einem neuen Fenster
Meeting-ID: 926 6433 1002
We look forward to seeing you there!