RandomerForest
RandomerForest copied to clipboard
tyler-tomita
Discriminant Projection Forest results, datasets, etc.
For simulations, we can analytically compute things and include this in pami
For benchmark datasets, plot: 1. Ensemble OOB error vs L (avg nonzeros per projection) and mtry, where line width indicates value of mtry. 2. Individual tree OOB errors vs individual...
plot it, compute it analytically simulate it
Randomly rotate the original spatial representation of the image. For each node, select a tile and find best split. Need to think about what the split criteria should be.
Focus on RerF being a robust rotationally invariant classifier Add rotation and node_mean experiments Compare RerF to other oblique tree classifiers (Breiman's Forest-RC) Add some theoretical results
Metadata
Owner
Metadata
Discriminant Projection Forest results, datasets, etc.