next up previous contents
Next: Supervised Self-Organising Map Up: Introduction Previous: Introduction

Definition of the Missing Values

Vectors x(t) are the potential data and vectors i(t) are vectors of same dimension as x(t). ik(t)=1 means xk(t) is observed, whereas ik(t)=0 means xk(t) is missing. i(t) are always observed. A typical problem in this setting is that x(t) should be reconstructed based on observed data xk(t)ik(t) and the indicator vectors i(t) [3].



Tapani Raiko
2001-09-26