Hi, so I read your github. But isnt this just vefy similar to a ann but with random functions and trees instead of activation functions and backwards propegation?
I would think that by creating random trees and hoping one performs well you're very unlikely to reach a global optimum or even a local optimum. So how does the performance and training time compare to traditional methods?
Rubscrub11 karma
Hi, so I read your github. But isnt this just vefy similar to a ann but with random functions and trees instead of activation functions and backwards propegation?
I would think that by creating random trees and hoping one performs well you're very unlikely to reach a global optimum or even a local optimum. So how does the performance and training time compare to traditional methods?
View HistoryShare Link