2000 character limit reached
Optical Fiber Communication Systems Based on End-to-End Deep Learning
Published 18 May 2020 in eess.SP, cs.IT, cs.LG, and math.IT | (2005.08785v1)
Abstract: We investigate end-to-end optimized optical transmission systems based on feedforward or bidirectional recurrent neural networks (BRNN) and deep learning. In particular, we report the first experimental demonstration of a BRNN auto-encoder, highlighting the performance improvement achieved with recurrent processing for communication over dispersive nonlinear channels.
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.
Top Community Prompts
Collections
Sign up for free to add this paper to one or more collections.
