Abstract-Potentiometry with ion-selective electrodes (ISEs) provides a simple and cheap approach for estimating ionic activities. However, a well-known shortcoming of ISEs regards their lack of selectivity. Recent works have suggested that smart sensor arrays equipped with a blind source separation (BSS) algorithm offer a promising solution to the interference problem. In fact, the use of blind methods eases the time-demanding calibration stages needed in the typical approaches. In this work, we develop a Bayesian source separation method for processing the outputs of an ISE array. The major benefit brought by the Bayesian framework is the possibility of taking into account some prior information, which can result in more realistic solutions. Concerning the inference stage, it is conducted by means of Markov chain Monte Carlo (MCMC) methods. The validity of our approach is supported by experiments with artificial data and also in a scenario with real data.

No abstract

Abstract. In this work, we address the problem of source separation of post-nonlinear mixtures based on mutual information minimization. There are two main problems related to the training of separating systems in this case: the requirement of entropy estimation and the risk of local convergence. In order to overcome both difficulties, we propose a training paradigm based on entropy estimation through order statistics and on an evolutionary-based learning algorithm. Simulation results indicate the validity of the novel approach.