Valentin Schmutz

This paper shows that heterogeneous spiking networks can produce fast rate-based dynamics without assuming Poisson spikes or slow dynamics.

This paper connects cortical circuit models with low-dimensional population dynamics using a recurrent spiking network model and demonstrates a method to retrieve circuit structure from recordings.

This paper uses AI to discover an equation explaining how neuronal tuning smoothness controls population code geometry in visual cortex and head-direction cells.

The paper presents a solvable RNN model showing low-dimensional latent dynamics can generate high-dimensional neuronal activity, and introduces NCE to demonstrate this in mouse visual cortex data.