2019/03/07 by Rick Fritschek, Fritschek, Rick, Rafael F. Schaefer +3
Computer Science · #Adversarial Robustness in Machine Learning #FOS: Computer and information sciences #Information Theory (cs.IT) #Machine Learning (cs.LG) #Speech and Audio Processing #Wireless Signal Modulation Classification
paper · pdf · doi:10.48550/arxiv.1903.02865
openalex publication_date 2019/03/07 · openalex created_date 2022/07/29 · openalex updated_date 2026/07/28
End-to-end deep learning for communication systems, i.e., systems whose\nencoder and decoder are learned, has attracted significant interest recently,\ndue to its performance which comes close to well-developed classical\nencoder-decoder designs. However, one of the drawbacks of current learning\napproaches is that a differentiable channel model is needed for the training of\nthe underlying neural networks. In real-world scenarios, such a channel model\nis hardly available and often the channel density is not even known at all.\nSome works, therefore, focus on a generative approach, i.e., generating the\nchannel from samples, or rely on reinforcement learning to circumvent this\nproblem. We present a novel approach which utilizes a recently proposed neural\nestimator of mutual information. We use this estimator to optimize the encoder\nfor a maximized mutual information, only relying on channel samples. Moreover,\nwe show that our approach achieves the same performance as state-of-the-art\nend-to-end learning with perfect channel model knowledge.\n