Risk-sensitive optimal control

摘要: Part 1 LQG theory summarized: structure, certainty equivalence the Markov case - control rules state estimation complete dualization of past and future. 2 Risk sensitivity, LEQG formulation: risk-sensitive certainty-equivalence principle future stress infinite horizon limits, break-down points policy improvement. 3 The path-integral (Hamiltonian) approach: methods, formalism recursions factorizations higher-order models general canonical factorization in context recoupling continuous time optimization. 4 Connections variations: relationship criterion to H entropy criteria variants.