Skip to main content
. 2021 May 13;15:626380. doi: 10.3389/fnbot.2021.626380

Table 1.

The number of phrases and the structure of the training/test sets, to validate the generalisation capabilities of the cognitive architecture at run-time.

Sample training phrases No. of training phrases Sample test phrases No. of test phrases
Objects (animals) _dog dog
_frog frog
_snake snake
3 _bee bee
_elephant elephant
10
Task 1 What is this _dog? 1 What is this _bee? 10
What kind of animal is this _dog? 1 What kind of animal is this _bee? 10
Task 2 Which one is the dog _dog _frog?
Which one is the frog _dog _frog?
4 Which one is the bee _bee _bear? P(10, 2) = 180
Which is a mammal _dog _frog?
Which is an amphibian _dog _frog?
4 Which is a mammal _bee _bear?
Which is an insect _bee _bear?
P(10, 2) = 180
Task 3 There are a dog, a frog and a snake
You see only _dog _frog
Which animal is missing?
6 There are a bee, a bug and a whale
You see only _whale _bug
Which animal is missing?
C(10, 2)·C(3, 2) = 360
TOTAL 19 750

Sample training and test phrases are given.