Parallel cascade prediction of turbulence induced wavefront tilt

Neural networks have been shown to be capable of predicting turbulence induced wavefront distortions to a degree useful for an adaptive optics system. The authors demonstrate on data from a laboratory simulation of imaging through turbulence that parallel cascades can predict wavefront tilts with accuracy equal to that reported for a neural network on the same data. However the parallel cascades train more than 50 times faster than backpropagation neural networks to achieve the same accuracy in forecasting wavefront tilt.