This project investigated the effectiveness of using the connectivity structure of real biological connectomes (e.g., Drosophila) as interchangeable reservoir layers in SoTA neural architectures like UNets, LSTMs, and Transformers.

Ensemble machine learning with reservoir neural networks

Here

Attention-based brain emulation neural netowrks

Here