DisCo-DSO: Coupling Discrete and Continuous Optimization for Efficient Generative Design in Hybrid Spaces
This addresses the problem of efficient generative design in hybrid spaces for applications like decision tree learning and symbolic regression, representing an incremental improvement over decoupled optimization methods.
The paper tackles black-box optimization in hybrid discrete-continuous and variable-length spaces, such as in decision tree learning and symbolic regression, by proposing DisCo-DSO, a joint optimization approach that uses a generative model to learn a distribution over design variables, resulting in fewer objective function evaluations and improved performance and sample efficiency, with demonstrated superiority over state-of-the-art methods in interpretable reinforcement learning with decision trees.
We consider the challenge of black-box optimization within hybrid discrete-continuous and variable-length spaces, a problem that arises in various applications, such as decision tree learning and symbolic regression. We propose DisCo-DSO (Discrete-Continuous Deep Symbolic Optimization), a novel approach that uses a generative model to learn a joint distribution over discrete and continuous design variables to sample new hybrid designs. In contrast to standard decoupled approaches, in which the discrete and continuous variables are optimized separately, our joint optimization approach uses fewer objective function evaluations, is robust against non-differentiable objectives, and learns from prior samples to guide the search, leading to significant improvement in performance and sample efficiency. Our experiments on a diverse set of optimization tasks demonstrate that the advantages of DisCo-DSO become increasingly evident as the complexity of the problem increases. In particular, we illustrate DisCo-DSO's superiority over the state-of-the-art methods for interpretable reinforcement learning with decision trees.