Extended Data Table 1: Final performance of Complex STDP approximation and Double T-Maze experiment.
From: Network of evolvable neural units can learn synaptic learning rules and spiking dynamics

Shows the mean and standard deviation of the performance in each environment over 30 trial runs. The ENU consistently outperforms the compared models (p<0.005). On the Double T-Maze Experiment, the LSTM and GRU model perform similarly (p=0.012), which was also observed in previous experiments. See also Fig. 5 for a more detailed convergence analysis.