Online semi-supervised learning: Application to dynamic learning from RADAR data

Dynamic learning from RADAR data is a new challenge which needs the development of new classification methods using an online semi-supervised learning approach. Except a very recent paper presented at ECML 2008, no algorithm in the literature can deal with this problem. Based on a theoretical analysis of the limitations and advantages of several semi-supervised and online learning methods, we proposed four algorithms as potential solutions to the dynamic learning problem which will be tested in next monthes.