Skip to main content

Erklärbare und sichere KI

Summer

(engl. Explainable and safe AI )

Modulnummer: FIN-INF-110496
Link zum LSF: LSF
Verantwortung: Sebastian Stober
Dozent:in: Sebastian Stober
Lehrveranstaltungen:
  • Vorlesung Erklärbare und sichere KI
  • Übung Erklärbare und sichere KI
Verwendbarkeit: - B.Sc. INF: Informatik - Wahlpflicht
- B.Sc. INF: Studienprofil: Künstliche Intelligenz
- B.Sc. CV: Informatik - Wahlpflicht
- B.Sc. INGINF: Informatik - Wahlpflicht
- B.Sc. WIF: Gestalten und Anwenden - Wahlpflicht
- B.Sc. INF (bilingual): Informatik - Wahlpflicht

Kürzel

XAI

CP

5

Semester

Summer

Fachsem.

ab 3.

Dauer

1 Semester

Sprache

deutsch

Niveau

Bachelor

Angestrebte Lernergebnisse:
The students ...

  • can confidently apply methods to explain decisions and internal representations, as well as processes of machine learning models
  • have the ability to assess the quality of models beyond accuracy metrics
  • can improve models in a targeted manner

Inhalt:
Classification of different categories of explainability methods for machine learning models and requirements for user-centered explanations. Presentation of directly interpretable models (ante-hoc) as well as analysis and visualization methods for black box models (post-hoc). In addition, basic post-hoc explanation techniques with model-agnostic and model-specific approaches are presented.

Arbeitsaufwand:

  • 56h attendance time (lecture + exercise)
  • 94h independent work (preparation and post-processing of lecture (OER) and exercise, working on exercise and programming tasks)

Prüfungsvorleistungen: Studien-/Prüfungsleistungen: Lehrform / SWS:

Written exam 120 minutes Announcement of the necessary preliminary work in the first week of the course.

  • Lecture (2 SWS)
  • Exercise (2 SWS)

Voraussetzungen nach Prüfungsordnung: Empfohlene Voraussetzungen:

none

Deep Learning for Engineers

Medienformen: Literatur:



Hinweise: