On some performances of the exponentiated convex variable step-size (ECVSS) algorithm

Recently, there were proposed few stochastic gradient algorithms which are based on cost functions that have exponential dependence on the chosen error. However, in certain framework the convergence of the cost function based on exponential of the squared error is not always satisfactorily. Thus this cost function has been modified and the Exponentiated Convex Variable Step-Size (ECVSS) algorithm has been obtained. It has been shown that with an optimal selection of one tuning parameter the ECVSS algorithm provides attractive results. In this brief we present some performances of the ECVSS stochastic gradient algorithm (time constants, misadjustment, computational complexity) in data echo cancellation framework when the tuning parameter has different values.