4 ms·
The easiest thing to do is probably use A* to find the solution, then imitation learning on the NN to learn it. (The immediate feedback gets rid of the vanishin
by rhaps0dy 1y ago
The easiest thing to do is probably use A* to find the solution, then imitation learning on the NN to learn it. (The immediate feedback gets rid of the vanishing/exploding gradients problem).