Demos for training recurrent spiking networks, as described in The centrality of population-level factors to network computation is demonstrated by a versatile approach for training spiking networks.
Demos were tested and successfully executed 05-15-2022 on MATLAB 2016a&b and MATLAB 2021b.
Train a spiking network to generate the cycling factors. The cycling factors and emg outputs are provided.
Train a spikng network to perform the reaching task. The reaching emg outputs are provided, and code is provided to derive the factors from a rate-based model.
Train a spiking network to perform the contextual integration task. A backprop trained rate network for generating the factors is already provided.
Train a spiking network to perform the cycling task and the reaching task. The cycling factors and emg outputs are provided. The reaching emg outputs are provided, and code is provided to derive the factors from a rate-based model.
Example of computing the factor based rate using cycling task factors.