Abstract
This paper concerns the problem of reduction of stochastic differential equations of a special type to ordinary differential equations parametrized by ω∊Ω New representations for the optimal filtration estimate of a conditional Gaussian process are obtained.