rustneat
rustneat copied to clipboard
Some concerns about our neural networks and Ctrnn
Organism
and Gene
- In
mutation_add_connection
, there is no check whether the connection already exists. InOrganism::get_weights
, we only use the weight of one connection, so any other similar connections are useless. - Bias can only be positive. I'm not sure if this is good? I also noticed in
Ctrnn
that we take the negative bias, rather than the positive as in the paper. So practically we always have a negative bias. - What about having both
bias: f64
andweight: f64
inGene
(instead of having a bias and a weight type of gene)?
Ctrnn
- Shouldn't tau be bigger than
1.0
? I think it sort of represents how fasty
decays. And when it is1.0
,y
decays in only one timestep. (you can verify this by looking at how we calculate the new y (however look at my commit where I fixed a small thing in Ctrnn)