Towards a harmonic blending of deterministic and stochastic frameworks in information processing

System Identification with Quantized Observations

“…Due to the coexistence of deterministic uncertainty from unmodeled dynamics and stochastic disturbances, we are treating necessarily a mixed environment. Consequently, estimation error characterization has a probabilistic measure that is compounded with a worst-case scenario over unmodeled dynamics, an idea introduced in our earlier work [101,102].…”

“…The following framework was introduced in [101,102]. It treats unmodeled dynamics as an unknown-but-bounded uncertainty.…”

“…The main focus of this chapter is on the derivation of the upper and lower bounds of estimation errors, which were also obtained in [111]. The framework for combining stochastic and deterministic approaches for identification was introduced in [101,102]. For some related but different identification algorithms, we refer the reader to [2,13,18,27,29,33,38,73,96,119] and references therein.…”