Semantic Interpolation in Implicit Models
This addresses a specific issue in generative modeling for researchers and practitioners, offering an incremental improvement to interpolation methods.
The paper tackled the problem of semantic interpolation in implicit models by showing that mismatched distributional assumptions and geometry can degrade interpolation quality, and proposed modifying the prior code distribution to concentrate probability mass near the origin, resulting in improved visual quality and more meaningful interpolation paths in experiments on benchmark image datasets.
In implicit models, one often interpolates between sampled points in latent space. As we show in this paper, care needs to be taken to match-up the distributional assumptions on code vectors with the geometry of the interpolating paths. Otherwise, typical assumptions about the quality and semantics of in-between points may not be justified. Based on our analysis we propose to modify the prior code distribution to put significantly more probability mass closer to the origin. As a result, linear interpolation paths are not only shortest paths, but they are also guaranteed to pass through high-density regions, irrespective of the dimensionality of the latent space. Experiments on standard benchmark image datasets demonstrate clear visual improvements in the quality of the generated samples and exhibit more meaningful interpolation paths.