Background
Recurrent neural networks (RNNs) have been tremendously helpful for undera analysis standing the dynamics underlying neural computations. One limiting factor of RNNs for understanding the brain circuits, is their weak connection with more biologically plausible networks.
Project theme
In this project we will train conventional RNN and more biologically plausible counterparts, and then explore how we can relate their low dimensional dynamics of RNN to their counterparts in more plausible biological neural networks.
Skill we might need (N=necessary, D=desired, P=plus)
- (N) Have background (bachelor) in (computational) neuroscience, neuroscience, physics, mathematics, statistics, machine learning, psychology, and other related fields.
- (N) Comfortable with programming (preferably with Python)
- (P) Familiar with neural networks
- (P) Familiar with dynamical systems
Learning targets
- Learn to train RNNs and relevant toolboxes
- Learn dimentionality reduction techniques in neuroscience and machine learning
- Learn topics in cognitive computational neuroscience
Relevant literature
Interested?
If you are interested please fill this form, and select the corresponding project code. ] If you need further information please get in touch with the contact person noted below.
