In this paper, we derive a novel probabilistic model of boosting as a Product of Experts. We re-derive the boosting algorithm as a greedy incremental model selection procedure which ensures that addition of new experts to the ensemble does not decrease the likelihood of the data. These learning rules lead to a generic boosting algorithm - POE-Boost which turns out to be similar to the AdaBoost algorithm under certain assumptions on the expert probabilities. The paper then extends the POEBoost algorithm to POEBoost.CS which handles hypothesis that produce probabilistic predictions. This new algorithm is shown to have better generalization performance compared to other state of the art algorithms.
|Translated title of the contribution||Boosting as a Product of Experts|
|Title of host publication||Uncertainty in Artificial Intelligence|
|Publisher||Association for Uncertainty in Artificial Intelligence Press|
|Number of pages||8|
|Publication status||Published - 2011|
Bibliographical noteConference Proceedings/Title of Journal: Uncertainty in Artificial Intelligence. Proceedings of the Twenty-Seventh Conference
Conference Organiser: Peter Grünwald et al