all AI news
[Project] How to organise inputs when there are variable/multiple/dynamic number of objects being tracked?
I have a 2D square with points in it, each point representing a creature. Each creature has a field of vision (FOV; a cone stemming from it). If creature 2 enters the FOV of creature 1, I collect features of creature 2 and input it into the NN of creature 1. Let's say colour, distance between, speed of creature 2, direction, so 4 features. So creature 1 has an input shape of 1x4.
Creature 1 sees creature 2 whose features form Inputs -> hidden layers -> output (the direction, speed …!-->