We formulate a new family of bootstrap algorithms suitable for learning non-Boolean functions from data. Within the Algorithmic Inference framework, the key idea is to consider a population of functions that are compatible with the observed sample. We generate items of this population from standard random seeds and reverse seed probabilities on the items. In this way we may compute in principle, and effectively achieve on paradigmatic examples, direct estimates and confidence intervals for any kind of complex function underlying the observed data according to any hypothesis on the randomness affecting the sample.
展开▼
