University of Bamberg
open search form
You are here

Information Systems and Applied Computer Sciences

Cognitive Systems

  1.  University of bamberg
  2. Faculties
  3. Information Systems and Applied Computer Sciences
  4. Subject groups
  5. Applied Computer Science
  6. Cognitive Systems
  7. Studies
  8. Courses
  9. Seminar Cognitive Systems

Masterseminar Kognitive Systeme/Cognitive Systems (SS 2025)

General Information

  • This seminar is open only for students of the bachelor programs AI and SoSySc and master programs CitH and AI.
  • As a bachelor student you have to pass our course AI-KI-B, as a master student you have to pass either AI-KI-B or KogSys-ML-B/M succesfully to be able to participate in the seminar.
  • There will be a limit of 25 students that are able to participate in the seminar.
  • You have to apply for this seminar in a central application procedure. More info in the respective VC course.
  • You find administrative information at UnivIS.
  • Participants should sign up for the course in the virtual campus.
  • The course is usually offered in the summer term.

Topic:

There are two large families of approaches in artificial intelligence - knowledge-based and data-driven approaches. The latter include deep neural networks, for example. These require large amounts of data and generate barely comprehensible (implicit/black box) machine-learned models. This entails a number of limitations, in particular high demands on the quantity and quality of data, high annotation effort, undesirable biases, susceptibility to anomalies and a lack of traceability. Alternatively, there are interpretable (explicit/whitebox) machine learning approaches, such as decision trees and inductive logic programming. These benefit from explicit, symbolic representations as well as logic and probability-based calculations that enable knowledge-based conclusions. Symbolic approaches can help to deal with the limitations of neural approaches and have properties such as explainability, robustness, adaptability, generalizability, abstraction, modelling of common sense and causality. However, the need to explicitly describe features, knowledge and rules can also be seen as a disadvantage compared to the end-to-end learning of neural networks from raw data.

Current research in the field of neuro-symbolic artificial intelligence (neuro-symbolic AI), which is developing hybrid approaches, is therefore becoming increasingly important. In particular, various ways of combining knowledge-based and data-driven approaches are being researched. Neuro-symbolic AI also includes research into measuring the plausibility and usefulness of hybrid approaches based on cognitive science findings. It is thus one of the central research topics for building human-centered and safe intelligent systems (e.g. in medicine, environmental protection, mobility or industrial production).

The seminar deals with theories and methods that combine the neural and symbolic worlds. Specific topics include: hybrid architectures and neuro-symbolic integration, symbolic representation (logic, ontologies and knowledge graphs), logical reasoning (inference, theorem proving and planning on symbolic representations), information interpretation, concepts and mental models, and explainability. In the summer semester 2024, the seminar will focus in particular on how human knowledge can be extracted from or integrated into machine-learned models.

Recommended Reading / Links / Topics

  • Hitzler, P., & Sarker, M. (2022). Neuro-Symbolic AI= Neural+ Logical+ Probabilistic AI. Neuro-Symbolic Artificial Intelligence: The State of the Art, 342, 173.
  • De Raedt, L., Duman?i?, S., Manhaeve, R., & Marra, G. (2021). From statistical relational to neural-symbolic artificial intelligence. In Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on ArtificialIntelligence (pp. 4943-4950).
  • Wagner, B., & d'Avila Garcez, A. S. (2021). Neural-symbolic integration for fairness in AI. In CEUR Workshop Proceedings (Vol. 2846).
  • Evans, R., Hernández-Orallo, J., Welbl, J., Kohli, P., & Sergot, M. (2021). Making sense of sensory input. Artificial Intelligence, 293, 103438.
  • Besold, T. R., d’Avila Garcez, A., Bader, S., Bowman, H., Domingos, P., Hitzler, P., ... & Zaverucha, G. (2021). Neural-Symbolic Learning and Reasoning: A Survey and Interpretation. In Neuro-SymbolicArtificialIntelligence: The State of the Art (pp. 1-51). IOS Press.
  • Skryagin, A., Stelzner, K., Molina, A., Ventola, F., Yu, Z., & Kersting, K. (2020). Sum-product logic: integrating probabilistic circuits into deepproblog. In Working Notes of the ICML 2020 Workshop on Bridge BetweenPerception and Reasoning: Graph Neural Networks and Beyond.
  • Mao, J., Gan, C., Kohli, P., Tenenbaum, J. B., & Wu, J. (2019). The neuro-symbolic concept learner: Interpreting scenes, words, and sentences from natural supervision. In International Conference on Learning Representations.
  • Evans, R., & Grefenstette, E. (2018). Learning explanatory rules from noisy data. Journal of Artificial Intelligence Research, 61, 1-64.

Previous Seminars


Bachelor/Master-Seminar: AI and Education:   [WS23/24]

Bachelor/Master-Seminar: Explainable Artificial Intelligence:  [WS 19/20] [WS 20/21] [WS22/23]

KI-Seminare (KI gestern, heute, morgen): [WS 15/16] [WS 16/17] [WS 17/18] [WS 18/19]

Bachelor Seminare: [WS 04/05] [WS 05/06] [WS 06/07] [WS 07/08] [WS 08/09] [WS 09/10]  [WS 10/11]    [SS 11] [WS 11/12] [WS 12/13] [WS13/14]

Master Seminare: [SS 05] [SS 06] [SS 08] [SS 09] [WS 09/10] [SS 10]  [WS 11/12] [WS 12/13] [WS 13/14] [SS 20] [SS 21][SS22] [SS23] [SS24]

Reading Clubs:

  • WS 14/15: Cognitive Models for Number Series Induction Problems  [Archiv Page]
  • SS 2014: Experimenting with a Humanoid Robot - Programming NAO to (Inter-)Act  [Archiv Page]
  • SS 2013: An introduction into statistic data analysis with R  [Archiv Page]
  • SS 2012: Transfer Learning  [Archiv Page]
  • SS 2011: Emotion Mining in Images and Text  [Archiv Page]
  • SS 2010: Aspects of Cognitive Robotics [Archiv Page]
  • SS 2009: Reading Club Decision Support Systems [Archiv Page]
  • WS 08/09: Algebraic Foundations of Functional Programming (together with Theoretical Computer Science) [Archiv Page]
  • SS 2008: Similarity (together with Statistics) [Archiv Page]
  • SS 2007: Automated Theorem Proving with Isabelle (together with Theoretical Computer Science) [Archiv Page]
  • SS 2006: Support Vector Machines [Archiv Page]