Meta and Virginia Tech's ReGDiff boosts novel metamaterial generation at NeurIPS
jiqizhixin · x · 2026-11-02
Virginia Tech and Meta researchers introduce ReGDiff, a NeurIPS 2026 paper on voxel-based metamaterial geometry generation.
- Problem: staying close to training samples limits novelty, while pushing too far breaks symmetry, periodicity, or connectivity.
- Method: a Repel-And-Sink (RAS) regularization separates valid structures from corrupted examples while spreading latent representations without unbounded drift; Short-Range Repulsion during diffusion sampling nudges candidates away from nearby known designs, with rapidly weakening influence at longer distances.
- Results: novelty reaches 0.296 on MetaTruss vs 0.163 for the Yang baseline, while preserving geometric plausibility.
More from Research
- Deep RL book excerpt: the policy loss is fake — a surrogate with no physical meaning — burkov · 2026-11-02
- NBER Paper Proposes Coasian Stakeholder Theory: Defining Stakeholders by Shutdown Power — sebkrier · 2026-11-02
- FAST-MaP Protocol Brings Single-Nucleotide RNA Structure Data in Under a Week via Nanopore — anshulkundaje · 2026-11-02
- A Free arXiv Paper Collection to Help You Discover Worthwhile Reads — omarsar0 · 2026-11-02
- livgorton: mech interp is pre-paradigmatic; "stuck in old paradigm" is a dodge — livgorton · 2026-11-02
- VTuber stream explainer on IPD × NVIDIA protein design model RFD4-Proteina — AllThingsApx · 2026-11-02