SPREAD: Sampling-based Pareto front Refinement via Efficient Adaptive Diffusion
This addresses the problem of computing Pareto sets for conflicting objectives in optimization, particularly for large-scale and expensive problems, representing a novel method for a known bottleneck.
The paper tackled the challenge of efficient multi-objective optimization for large-scale problems by introducing SPREAD, a generative framework based on Denoising Diffusion Probabilistic Models, which matched or exceeded leading baselines in efficiency, scalability, and Pareto front coverage on benchmarks.
Developing efficient multi-objective optimization methods to compute the Pareto set of optimal compromises between conflicting objectives remains a key challenge, especially for large-scale and expensive problems. To bridge this gap, we introduce SPREAD, a generative framework based on Denoising Diffusion Probabilistic Models (DDPMs). SPREAD first learns a conditional diffusion process over points sampled from the decision space and then, at each reverse diffusion step, refines candidates via a sampling scheme that uses an adaptive multiple gradient descent-inspired update for fast convergence alongside a Gaussian RBF-based repulsion term for diversity. Empirical results on multi-objective optimization benchmarks, including offline and Bayesian surrogate-based settings, show that SPREAD matches or exceeds leading baselines in efficiency, scalability, and Pareto front coverage.