thx, it is a simple feedforward, 5 inputs (the rays), maybe 1 hidden layer with 4 neurons, and 4 outputs (turn right/left, brake/accelerate)
trained from a set of inputs pairs. As you play, it records a consequent number of pairs to feed in the annet, so that the annet "tries to immitate" your playing.
thx, it is a simple feedforward, 5 inputs (the rays), maybe 1 hidden layer with 4 neurons, and 4 outputs (turn right/left, brake/accelerate)
trained from a set of inputs pairs. As you play, it records a consequent number of pairs to feed in the annet, so that the annet "tries to immitate" your playing.
Very basic one... but it does the job !
labenjam 4 years ago