Here's much delayed post on my experience at Cosyne 2024!
Big themes at Cosyne:
1. Compositionality (in task representation, perception and planning)
2. Neural manifolds (low dimension structures in neural data while performing a task)
3. Using RNNs for standard neuro/psych tasks
But that's not all, there's many other interesting talks and posters (and workshop sessions) that I can't go through in blog, but all compiled in these slides I created for a lab presentation
What themes I think could be up and coming in future Cosynes -
1. Compositionality (again) and structure learning
2. Continual learning (especially in motor repertoires and other skills)
3. Solving complex tasks with local learning rules
4. Homeostatic RL
5. Neuro specific RL gym environments opposed to OpenAI's standard RL gym environments
Until next time :)
