Gradient-Free Training of Autoencoders for Non-Differentiable Communication Channels

Training of autoencoders using the back-propagation algorithm is challenging for non-differential channel models or in an experimental environment where gradients cannot be computed. In this paper, we study a gradient–free training method based on cubature Kalman filter. To numerically validate method, autoencoder employed to perform geometric constellation shaping differentiable communication channels, showing same performance as algorithm. Further investigation done non–differentiable that includes: laser phase noise, additive white Gaussian noise and blind search-based compensation. Our results indicate can successfully optimized proposed achieve better robustness residual with respect standard schemes such Quadrature Amplitude Modulation Iterative Polar considered conditions.