Parameterless-Growing-SOM and Its Application to a Voice Instruction Learning System

Table 2

Parameters used in the experiments.

Description	Symbol	Quantity
Size of Feature Map of T-SOM	$N\times M$	$5\times 5$
Size of initial PL-G-SOM	$N\times M$	$5\times 5$
Iteration times for one instruction	$t$	$300\times 4$
Temperature	$T$	1.0
Number of instructions (actions)	$a(i)$	4
Maximum/Minimum neighborhood in PL-G-SOM	${\sigma }_{\max \hspace{0.17em}},$${\sigma }_{\min \hspace{0.17em}}$	$N\times M/2,\hspace{0.17em}\hspace{0.17em}0.7$
Reward for one action selected	$r$	10.0
Parameters of Feeling Map	$a,b$	0.2, 0.05
Number of samples	—	48
Sampling rate	—	8 KHz
Sample size	—	8 bit
Channel	—	monaural