moospread#
The moospread package implements in PyTorch the SPREAD method proposed in our paper
SPREAD: Sampling-based Pareto Front Refinement via Efficient Adaptive Diffusion.
SPREAD is a sampling-based approach for multi-objective optimization that leverages diffusion models to refine and generate well-spread Pareto front approximations efficiently. It combines the expressive power of diffusion models with multi-objective optimization principles to achieve both strong convergence to the Pareto front and high diversity across the objective space.
|
|
|