LAMP: Data-Efficient Linear Affine Weight-Space Models for Parameter-Controlled 3D Shape Generation and Extrapolation
Researchers have developed LAMP, a new framework for generating 3D shapes with precise parameter control. This method is highly data-efficient, requiring as few as 50 samples to achieve controlled interpolation and safe extrapolation. LAMP also includes a safety metric to detect when mixed decoder outputs might be unreliable, outperforming existing methods in data efficiency and parameter fidelity. AI
IMPACT Enables more efficient and reliable 3D design exploration and generation with limited data.