Reproduce “DnNP: Denoising Input Uncertainty in Neural Processes” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Fatemeh Tohidian, Chengzhi Shi, Soomi Lee, Matthew Elia, Amy Mueller, Stratis Ioannidis, Jennifer Dy
The authors introduce DnNP, a Neural Process that models noisy inputs by estimating a distribution over clean inputs and using them for prediction. This adds measurement uncertainty to uncertainty about the function and output. On Brookline PM₂.₅ interpolation with latitude, longitude, and relative humidity, DnNP lowers RMSE about 30% and NLL about 73% versus NP, but evidence is dataset-specific.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.