Scalable Semi-Supervised Aggregation of Classifiers
This addresses the challenge of enhancing classifier ensemble accuracy in a scalable, assumption-free manner, though it appears incremental as it builds on existing aggregation methods.
The paper tackles the problem of aggregating predictions from an ensemble of binary classifiers by using unlabeled data to improve performance, achieving significant gains as demonstrated empirically with random forests.
We present and empirically evaluate an efficient algorithm that learns to aggregate the predictions of an ensemble of binary classifiers. The algorithm uses the structure of the ensemble predictions on unlabeled data to yield significant performance improvements. It does this without making assumptions on the structure or origin of the ensemble, without parameters, and as scalably as linear learning. We empirically demonstrate these performance gains with random forests.