Alert button

Active Nearest Neighbor Regression Through Delaunay Refinement

Jun 16, 2022
Alexander Kravberg, Giovanni Luca Marchetti, Vladislav Polianskii, Anastasiia Varava, Florian T. Pokorny, Danica Kragic

Figure 1 for Active Nearest Neighbor Regression Through Delaunay Refinement
Figure 2 for Active Nearest Neighbor Regression Through Delaunay Refinement
Figure 3 for Active Nearest Neighbor Regression Through Delaunay Refinement
Figure 4 for Active Nearest Neighbor Regression Through Delaunay Refinement

Share this with someone who'll enjoy it:

We introduce an algorithm for active function approximation based on nearest neighbor regression. Our Active Nearest Neighbor Regressor (ANNR) relies on the Voronoi-Delaunay framework from computational geometry to subdivide the space into cells with constant estimated function value and select novel query points in a way that takes the geometry of the function graph into account. We consider the recent state-of-the-art active function approximator called DEFER, which is based on incremental rectangular partitioning of the space, as the main baseline. The ANNR addresses a number of limitations that arise from the space subdivision strategy used in DEFER. We provide a computationally efficient implementation of our method, as well as theoretical halting guarantees. Empirical results show that ANNR outperforms the baseline for both closed-form functions and real-world examples, such as gravitational wave parameter inference and exploration of the latent space of a generative model.

* Accepted at the International Conference on Machine Learning (ICML) 2022  
View paper onarxiv icon

Share this with someone who'll enjoy it: