Cognitive maps in natural and artificial intelligence (MTAT.03.292)
Topic: Cognitive maps in neuroscience and their influence in computer science.
Course ÕIS page
- Seminars: 9 double contact lessons on Thursdays from 14:15 to 16:15
- Location: Delta room 2006
- Questions: Jesús Javier Reyes Torres, Francisco José Maldonado Torralba
- NAI lab: https://nail.cs.ut.ee/
Description
We welcome all doctorate, master's, and curious 3rd-year students interested in the frontier between neurosicence and computer science. This autumm we will cover the neural basis of representation of space, and how this drives innovations in artificial intelligence. We meet in person at the Delta Center in Tartu.
Schedule
Lecture | Topic | Resources |
12 Sep 2024 | Introduction | None |
26 Sep 2024 | Cognitive maps in neuroscience - Part 1 | What is a cognitive map? |
03 Oct 2024 | Cognitive maps in neuroscience - Part 2 | What is a cognitive map? |
10 Oct 2024 | Representation of abstract concepts | Geometry of abstract learned knowledge in the hippocampus |
17 Oct 2024 | Insights on concept learning | Minimization of Boolean complexity in human concept learning |
24 Oct 2024 | A quantitative approach to simplicity | Ockham's Razor and Bayesian Analysis |
31 Oct 2024 | Development of spatial encoding | Grid cells in rats deprived of geometric experience during development |
07 Nov 2024 | How we encode others | Social place-cells in the bat hippocampus |
14 Nov 2024 | Cognition from Bayesian analysis | How to Grow a Mind: Statistics, Structure, and Abstraction |
21 Nov 2024 | Artificial intelligence for navigation | Vector-based navigation using grid-like representations in artificial agents |
28 Nov 2024 | Properties of the spatial cognitive map | The entorhinal grid map is discretized |
05 Dec 2024 | Encoding for model generalization | DPP-A Over Grid Codes Supports Out of Distribution Generalization |
12 Dec 2024 | Wrap-up discussion session | Led by the lecturers |
Disclaimer: The timetable can be modified according to the assignation of workload during the first lecture.
Process
We use the flipped classroom method. Our seminars are discussion- and group-work-based, and students complete readings and assignments before the class. Each person must write a small paragraph at the end of each lecture with what they learned from the presentation and discussion, and if they have any remaining questions on the topic.
Grading
This is a pass/fail course, and no grades are given.
You can only miss one of the in-person seminars without justification to pass the course.