Richard Baraniuk, Randall Balestriero, Ahmed Imtiaz Humayun
We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. If it links a repository, it is listed below.
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
We present Polarity Sampling, a theoretically justified plugand-play method for controlling the generation quality and diversity of any pre-trained deep generative network (DGN). Leveraging the fact that DGNs are, or can be approximated by, continuous piecewise affine splines, we derive the analytical DGN output space distribution as a function of the product of the DGN's Jacobian singular values raised to a power ρ. We dub ρ the polarity parameter and prove that ρ focuses the DGN sampling on the modes (ρ < 0) or anti-modes (ρ > 0) of the DGN outputspace probability distribution. We demonstrate that nonzero polarity values achieve a better precision-recall (qualitydiversity) Pareto frontier than standard methods, such as truncation, for a number of state-of-the-art DGNs. We also present quantitative and qualitative results on the improvement of overall generation quality (e.g.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2203.01993")
get_code_for_paper("2203.01993")
have("2203.01993")
Connect an agent — have() is free.