Trustworthy Machine Learning
Modulnummer: Q06-16
Englischer Titel: Trustworthy Machine Learning
Leistungspunkte: 6
Lehrperson: Zehlike
Empfohlene Vorkenntnisse
Es werden allgemeine Kenntnisse in künstlicher Intelligenz und im maschinellen Lernen
vorausgesetzt, entweder durch einführende Veranstaltungen oder vergleichbare Kurse.
Zudem werden ein hohes Interesse an und die Fähigkeit zum Verstehen philosophischer
Essays vorausgesetzt.
Zwingende Voraussetzungen
keine
Inhalt
Accuracy is not enough when you’re developing machine learning systems for
consequential application domains. You also need to make sure that your models are fair,
have not been tampered with, will not fall apart in different conditions, and can be
understood by people. Your design and development process has to be transparent and
inclusive. You don’t want the systems you create to be harmful, but to help people flourish
in ways they consent to. All of these considerations beyond accuracy that make machine
learning safe, responsible, and worthy of our trust have been described by many experts
as the biggest challenge of the next five years. This course will equip you with the thought
process to meet this challenge.
The course focuses on three key issues in machine learning, addressed from an ethical,
legal, and technological perspective:
1. Personal data processing: privacy, confidentiality, surveillance, recourse, data
collection, and power differentials
2. Data-driven decision support: biases and transparency in data processing,
data-rich communication, and data visualization
3. Automated decision making: conceptualizations of power and discrimination in
scenarios with different degrees of automation
We will spend about half of the course studying computing technologies for, e.g.,
anonymizing data, or detecting and mitigating algorithmic bias. The other half of the
course we will study different conceptualizations of power around data processing
pipelines, analyze bias and discrimination in computer systems from a moral philosophy
perspective, and overview the relevant legal frameworks for data processing.
Format
The course will be derived through interactive online and in-person sessions.
Depending on the pandemic development, we might go fully remote.
- 15 theory sessions in which the main concepts, background, methods, criteria
will be introduced.
- 15 practice sessions in which case studies of current and envisioned scenarios
of data processing will be conducted in small groups and in which students will
present recent research papers on trustworthy machine learning.
Each student prepares a 20min talk with 10min Q&A.
Erforderliche Arbeitsleistungen für LP-Vergabe und Prüfungszulassung
Die Leistungen in der Übung (Gruppenarbeit, Präsentation) werden zur Prüfungszulassung
herangezogen. Details werden in der Veranstaltung bekannt gegeben.
Lehrveranstaltungen
Vorlesung: 2 SWS
Übung: 2 SWS
Zugeordneter Vertiefungsschwerpunkt
Algorithmen und Modelle: nein
Modellbasierte Systementwicklung: nein
Daten- und Wissensmanagement: ja
Ohne Vertiefungsschwerpunkt: nein
Sprache im Modul
Deutsch: nein
Englisch: ja
Angeboten für Studiengänge
M. Sc.: ja
M. Ed.: ja
Wirtschaftsmaster: ja
Angeboten im
Wintersemester: nein
Sommersemester: nein
Turnus
Unregelmäßig