This paper discusses the optimization of results derived from evolutionary algorithms by augmenting them with generative adversarial nets (GAN). The proposed research presents a hybrid algorithm that combines GAN with a genetic algorithm (GA). GAN are used for sampling the training data that is injected into the GA. The claim is that this approach helps GA find optimal solutions and avoid premature convergence, causing local optimization solutions when the complexity of the tasks increases.

The paper describes in detail the specifics of evolutionary algorithms and GAN, along with other approaches. It presents and explains the algorithm in such a way that it can be easily reproduced. Further, the experiments and test results compare the performance of the hybrid algorithm with the performance of the traditional GA, showing an improvement of almost 100 percent for some of the mean and min error rates. Finally, the conclusion section includes suggestions for improving the performance of the hybrid algorithm by further optimizing the GAN.

Very clearly written, the paper presents the proposed method with sufficient detail. It is a very good read for scholars, students, and practitioners interested in machine intelligence and learning.