OPT-GAN: Global Black-box Optimization by Learning Distribution of Optima

Black-box optimization (BBO) algorithms are concerned with finding the best solutions for problems with missing analytical details. Most classical methods for such problems are based on strong and fixed \emph{a priori} assumptions, such as Gaussian distribution. However, the complex real-world problems, especially when the global optimum is desired, could be very far from the \emph{a priori} assumptions because of their diversities, bringing some unexpected obstacles to these methods. In this paper, we present an optimizer using generative adversarial nets (OPT-GAN) to adapt to diverse black-box problems via estimating the distribution of optima. The method learns the extensive distribution of the optimal region dominated by selective and randomly moving candidates, balancing the exploration and exploitation. Experiments demonstrate that on BBOB problems and several other benchmarks with atypical distributions, OPT-GAN outperforms other classical BBO algorithms, in particular the ones with Gaussian assumptions.