Proximal Diffusion Neural Sampler
About
The task of learning a diffusion-based neural sampler for drawing samples from an unnormalized target distribution can be viewed as a stochastic optimal control problem on path measures. However, the training of neural samplers can be challenging when the target distribution is multimodal with significant barriers separating the modes, potentially leading to mode collapse. We propose a framework named Proximal Diffusion Neural Sampler (PDNS) that addresses these challenges by tackling the stochastic optimal control problem via proximal point method on the space of path measures. PDNS decomposes the learning process into a series of simpler subproblems that create a path gradually approaching the desired distribution. This staged procedure traces a progressively refined path to the desired distribution and promotes thorough exploration across modes. For a practical and efficient realization, we instantiate each proximal step with a proximal weighted denoising cross-entropy (WDCE) objective. We demonstrate the effectiveness and robustness of PDNS through extensive experiments on both continuous and discrete sampling tasks, including challenging scenarios in molecular dynamics and statistical physics. Our code is available at https://github.com/AlexandreGUO2001/PDNS.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| n-body particle system sampling | DW-4 d = 8 | W2 Distance0.51 | 29 | |
| Target Distribution Sampling | Funnel 10D | Sinkhorn Distance129.5 | 29 | |
| n-body particle system sampling | LJ-13 (d = 39) | W2 Distance1.57 | 21 | |
| Toy target distribution sampling | GMM40 d = 50 | W2 (Entropy Regulated, eps=0.05)327.8 | 18 | |
| n-body particle system sampling | LJ-55 d = 165 | W23.95 | 16 | |
| Learning Continuous Target Distributions | MoS d = 50 | Sinkhorn Cost353.1 | 11 | |
| Target Distribution Sampling | Many-Well 5D | Sinkhorn Distance0.08 | 11 | |
| Sampling from lattice Ising models | Lattice Ising model beta=0.28 L=24 (high-temperature) | Delta Mag.0.0039 | 6 | |
| Molecular Boltzmann Distribution Sampling | Alanine Dipeptide | KL Divergence (phi)0.02 | 5 | |
| Discrete sampling | Lattice Potts model L=16, N=4, beta=1.3 | Delta Mag.8.40e-4 | 4 |