The problem of making sense of data can be baptised as the data-to-information conversion step. Strong mathematical skills coupled with the inventive use of statistical models can uncover the sought-after information content in the sensed signal. Many times this data-to-inrofmation step includes:
- The bottom-to-top design of a signal model and the formulation of a meaningful statistical model.
- The simulation of the system with the expected input and with varying distributions of model parameters.
- Attempts for the quintification of the statistical error and to identify which part of it is attributed to the signal model discrepancies.
- The derivation of perfomance guarantees that govern the parameter inference procedure and optimisation algorithms.
- The application of the results in classification or regression.