Neural network-based anomalous diffusion parameter estimation approaches for Gaussian processes

Abstract Anomalous diffusion behavior can be observed in many single-particle (contained crowded environments) tracking experimental data. Numerous models used to describe such In this paper, we focus on two common processes: fractional Brownian motion (fBm) and scaled (sBm). We proposed novel methods for sBm anomalous parameter estimation based the autocovariance function (ACVF). Such a function, centered Gaussian processes, allows its unique identification. The first method is solely theoretical calculations, other one additionally utilizes neural networks (NN) achieve more robust well-performing estimator. Both fBm were compared between estimators ones utilizing artificial NN. For NN-based approaches, architectures as multilayer perceptron (MLP) long short-term memory (LSTM). Furthermore, analysis of additive noise influence estimators’ quality was conducted NN with without regularization method.