End-to-End Differentiable Diffusion Conditioning for Physics-Informed Optimization
Differentiable Diffusion Conditioning (D2C) is proposed for offline MBO, which leverages the fact that deterministic sampling makes the reverse denoising chain differentiable with respect to the conditioning signal and turns a frozen conditional diffusion model into an end-to-end gradient-based optimizer.
Ricardo Luna Gutiérrez, Vineet Gundecha, R. Ejaz et al.
· 0 citations