Optimum Engine Mounting Layout using MPOD

Facial pose estimation using modified synergetic computer

“…Considering the notion of penalty function from Fukushima et al [11], introducing the penalty terms in the learning of adjoint prototype matrix, up to certain extend, may generalize the inverse matrix and minimize the learning errors, hence, resolve overfitting issue. We named this modification as Modified SC-MELT.…”

“…The notion of Most Probable Optimum Design (MPOD) [11] has inspired the generalization for optimal design during learning phase. Hence, we worked out a novel learning function to replace the method proposed in original SC-MELT in which it is possible to reduce the learning error by performing generalization during the learning of the adjoint prototype.…”

“…Instead of solving (12) directly, we followed the direction of MPOD [11]. Firstly, let assume the relationship between learning prototype matrix V and the corresponding Kronecker delta function response stored in enhanced identity matrix E is coded by the adjoint prototype matrix V called synergetic memory and is expressed as:…”

“…The value of v: that makes the above equation minimum is [11]: (19) and if equation above is expressed in matrix form, we have:…”

“…In order to include the "melting" level adjustment and improvement over the adjoint prototype learning, this work follows the direction of Fukushima et al [11] in the introduction of the use of penalty function to generalize the inverse matrix. The adaption of MPOD penalty terms and modification yield the following equation:…”