An approach to decision making problems under statistical uncertainty

The paper proposes an approach to decision-making problems with the distribution of signal and noise probabilities unknown. The useful signal is assumed representable as a functional of conditional distribution of the observed signal. A special observation model is used.