Learning to reproduce fluctuating time series by inferring their time-dependent stochastic properties: Application in robot learning via tutoring,” Autonomous Mental Development

This study proposes a novel type of dynamic neural network model that can learn to extract stochastic or uctuating structures hidden in time series data. The network learns to predict not only the mean of the next input state, but also its time-dependent variance. The training method is based on maximum likelihood estimation by using the gradient descent method, and the likelihood function is expressed as a function of the estimated variance. Regarding the model evaluation, we present numerical experiments in which training data were generated in different ways of utilizing Gaussian noise. Our analysis showed that the network can predict the time-dependent variance and the mean, as well as that it can repro-duce the target stochastic sequence data by utilizing the estimated variance. Furthermore, it was shown that a humanoid robot using the proposed network can learn to reproduce la-tent stochastic structures hidden in uctuating tutoring trajectories. This learning scheme is essential for the acquisition of sensory-guided skilled behavior.