Feedback 12-03-2012

Tiger Problem

• I've changed my existing Tiger Problem environment to continuous observations 
• One-dimensional observation: -1.0 (tiger is left) to 1.0 (tiger is right)
• Observation is the position of the tiger corrupted by some zero-mean, normally distributed noise with a low standard deviation σ

Implementation Plan

1. General structures for nodes and trees (incorporating the shared interfaces)
2. Tree for observations [1,2]
3. Simulation to generate examples (for tree for observations)
4. Combine tree for observations with remainder of algorithm

[1] Elena Ikonomovska, Joao Gama: Learning Model Trees from Data Streams. Discovery Science 2008: 52-63.
[2] Geoff Hulten, Pedro Domingos, Laurie Spencer. Mining Massive Data Streams. Journal of Machine Learning Research (2005).
[3] P. Domingos and G. Hulten. Mining high-speed data streams. Proceedings of the sixth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining - KDD ’00 (2000) 71–80.3.