Table 3.

Hyper-parameter values unique to CTDL. ${\tau }_{\eta }$, ${\tau }_{\delta }$, $\sigma$, ${\sigma }_c$, $\alpha$ and $\rho$ were selected by using a random grid search on a single grid world.

Parameter	Value	Description
$U$	36	Number of units in SOM
${\tau }_{\eta }$	10	Temperature for calculating $\eta$
${\tau }_{\delta }$	1	Temperature for calculating $\delta$
$\sigma$	.1	Standard deviation of the SOM neighbourhood function
${\sigma }_c$	.1	Constant for denominator in SOM neighbourhood function
$\alpha$	.01	Learning rate for updating the weights of the SOM
$\rho$	.9	Learning rate for updating the Q values of the SOM