This is part of the Keras code for our paper: W. Xia, G. Zheng, Y. Zhu, J. Zhang, J. Wang, A. P. Petropulu, “A Deep Learning Framework for Optimization of MISO Downlink Beamforming,” IEEE Trans. Commun., vol. 68, no. 3, pp. 1866-1880, Mar. 2020.
It performs supervised training and then testing for the sum rate problem for a system with a 2-antenna base station and 2 users. The traing data is generated by using the weighted minimum mean squared error (WMMSE) algorithm, e.g., the one in the reference: S. S. Christensen, R. Agarwal, E. De Carvalho, and J. M. Cioffi, “Weighted sum-rate maximization using weighted MMSE for MIMO-BC beamforming design,” IEEE Trans. Wireless Commun., vol. 7, no. 12, 1069 pp. 4792–4799, Dec. 2008.