LAUSR

This paper makes an in depth study on the applications of distributed machine learning based techniques for parameter estimation of infinite impulse response (IIR) systems and as well as inverse modeling of nonlinear systems or sensors. The bio-inspired learning algorithms such as particle swarm optimization (PSO) and differential evolution (DE) are used as incremental and diffusion based distributed learning strategies to estimate the pole-zero parameters of a feed forward-feedback systems. The same distributed learning algorithms are also employed to generate inverse model of nonlinear systems. The performance of these learning algorithms in terms of accuracy of estimation are compared under different additive noise conditions. The ranking based on accuracy of direct estimation demonstrates that the proposed Incremental DE (IDE) based model performs the best than Diffusion DE (DDE) counter part. It is then followed by IPSO, DPSO, ILMS and DLMS based models. The same ranking is also valid for inverse modeling problem. The proposed distributed bioinspired learning can also be applied to various forecasting and classification tasks.