Extremely and very low frequency (ELF/VLF) radio waves are generated from a variety of natural geophysical sources. Ground‐based observations often contain signals of interest; however, the signals are typically immersed in a noisy environment due to lightning‐generated sferics and additional anthropogenic sources. Although automated detection algorithms have been employed successfully in the past, extraction of arbitrary and broadband signal classes has been a challenge. In this work, we employ a mask‐scoring regional convolutional neural network (MSRCNN) for automated extraction of whistlers from ground measurements at Palmer station, Antarctica. Statistics of several hundred thousand whistler receptions are evaluated to determine seasonal and diurnal variations at Palmer station along with strong correlations to lightning activity in the conjugate hemisphere. Although MSRCNN has been employed for whistler extraction in this work, the method has can be easily extended to other signal classes including chorus, hiss, and VLF triggered emissions.
The echo state network (ESN) is a special type of recurrent neural networks for processing the time-series dataset. However, limited by the strong correlation among sequential samples of the agent, ESN-based policy control algorithms are difficult to use the recursive least squares (RLS) algorithm to update the ESN's parameters. To solve this problem, we propose two novel policy control algorithms, ESNRLS-Q and ESNRLS-Sarsa. Firstly, to reduce the correlation of training samples, we use the leaky integrator ESN and the mini-batch learning mode. Secondly, to make RLS suitable for training ESN in mini-batch mode, we present a new mean-approximation method for updating the RLS correlation matrix. Thirdly, to prevent ESN from over-fitting, we use the L1 regularization technique. Lastly, to prevent the target state-action value from overestimation, we employ the Mellowmax method. Simulation results show that our algorithms have good convergence performance.