Learning from dependent aggregate observations
Defense Date:
The aim of the thesis is to extend the range of applications of a method of learning from aggregate observations [1]. A method of learning from dependent aggregates is introduced. The examined case assumes dependence of the aggregates in form of feature values commonality in aggregated elements. Experiments consist of verification of the method of learning from aggregate observations, examination of the influence of observation dependencies on learning capabilities, introduction of the method for solving the problem of aggregate interdependence along with confirmation of its effectiveness. The results indicated that the proposed method allows learning from dependent observations.